OLD | NEW |
| (Empty) |
1 // Copyright (c) 2012 The Chromium Authors. All rights reserved. | |
2 // Use of this source code is governed by a BSD-style license that can be | |
3 // found in the LICENSE file. | |
4 | |
5 #include "components/autofill/renderer/form_autofill_util.h" | |
6 | |
7 #include <map> | |
8 | |
9 #include "base/command_line.h" | |
10 #include "base/logging.h" | |
11 #include "base/memory/scoped_vector.h" | |
12 #include "base/metrics/field_trial.h" | |
13 #include "base/string_util.h" | |
14 #include "base/utf_string_conversions.h" | |
15 #include "components/autofill/common/autofill_switches.h" | |
16 #include "components/autofill/common/form_data.h" | |
17 #include "components/autofill/common/form_field_data.h" | |
18 #include "components/autofill/common/web_element_descriptor.h" | |
19 #include "third_party/WebKit/public/platform/WebString.h" | |
20 #include "third_party/WebKit/public/platform/WebVector.h" | |
21 #include "third_party/WebKit/Source/WebKit/chromium/public/WebDocument.h" | |
22 #include "third_party/WebKit/Source/WebKit/chromium/public/WebElement.h" | |
23 #include "third_party/WebKit/Source/WebKit/chromium/public/WebExceptionCode.h" | |
24 #include "third_party/WebKit/Source/WebKit/chromium/public/WebFormControlElement
.h" | |
25 #include "third_party/WebKit/Source/WebKit/chromium/public/WebFormElement.h" | |
26 #include "third_party/WebKit/Source/WebKit/chromium/public/WebFrame.h" | |
27 #include "third_party/WebKit/Source/WebKit/chromium/public/WebInputElement.h" | |
28 #include "third_party/WebKit/Source/WebKit/chromium/public/WebLabelElement.h" | |
29 #include "third_party/WebKit/Source/WebKit/chromium/public/WebNode.h" | |
30 #include "third_party/WebKit/Source/WebKit/chromium/public/WebNodeList.h" | |
31 #include "third_party/WebKit/Source/WebKit/chromium/public/WebOptionElement.h" | |
32 #include "third_party/WebKit/Source/WebKit/chromium/public/WebSelectElement.h" | |
33 | |
34 using WebKit::WebDocument; | |
35 using WebKit::WebElement; | |
36 using WebKit::WebExceptionCode; | |
37 using WebKit::WebFormControlElement; | |
38 using WebKit::WebFormElement; | |
39 using WebKit::WebFrame; | |
40 using WebKit::WebInputElement; | |
41 using WebKit::WebLabelElement; | |
42 using WebKit::WebNode; | |
43 using WebKit::WebNodeList; | |
44 using WebKit::WebOptionElement; | |
45 using WebKit::WebSelectElement; | |
46 using WebKit::WebString; | |
47 using WebKit::WebVector; | |
48 | |
49 namespace autofill { | |
50 namespace { | |
51 | |
52 // The maximum length allowed for form data. | |
53 const size_t kMaxDataLength = 1024; | |
54 | |
55 bool IsOptionElement(const WebElement& element) { | |
56 CR_DEFINE_STATIC_LOCAL(WebString, kOption, ("option")); | |
57 return element.hasTagName(kOption); | |
58 } | |
59 | |
60 bool IsScriptElement(const WebElement& element) { | |
61 CR_DEFINE_STATIC_LOCAL(WebString, kScript, ("script")); | |
62 return element.hasTagName(kScript); | |
63 } | |
64 | |
65 bool IsNoScriptElement(const WebElement& element) { | |
66 CR_DEFINE_STATIC_LOCAL(WebString, kNoScript, ("noscript")); | |
67 return element.hasTagName(kNoScript); | |
68 } | |
69 | |
70 bool HasTagName(const WebNode& node, const WebKit::WebString& tag) { | |
71 return node.isElementNode() && node.toConst<WebElement>().hasTagName(tag); | |
72 } | |
73 | |
74 bool IsAutofillableElement(const WebFormControlElement& element) { | |
75 const WebInputElement* input_element = toWebInputElement(&element); | |
76 return IsAutofillableInputElement(input_element) || IsSelectElement(element); | |
77 } | |
78 | |
79 bool IsAutocheckoutEnabled() { | |
80 return base::FieldTrialList::FindFullName("Autocheckout") == "Yes" || | |
81 CommandLine::ForCurrentProcess()->HasSwitch( | |
82 switches::kEnableExperimentalFormFilling); | |
83 } | |
84 | |
85 // Check whether the given field satisfies the REQUIRE_AUTOCOMPLETE requirement. | |
86 // When Autocheckout is enabled, this requirement is enforced in the browser | |
87 // process rather than in the renderer process, and hence all fields are | |
88 // considered to satisfy this requirement. | |
89 bool SatisfiesRequireAutocomplete(const WebInputElement& input_element) { | |
90 return input_element.autoComplete() || IsAutocheckoutEnabled(); | |
91 } | |
92 | |
93 // Appends |suffix| to |prefix| so that any intermediary whitespace is collapsed | |
94 // to a single space. If |force_whitespace| is true, then the resulting string | |
95 // is guaranteed to have a space between |prefix| and |suffix|. Otherwise, the | |
96 // result includes a space only if |prefix| has trailing whitespace or |suffix| | |
97 // has leading whitespace. | |
98 // A few examples: | |
99 // * CombineAndCollapseWhitespace("foo", "bar", false) -> "foobar" | |
100 // * CombineAndCollapseWhitespace("foo", "bar", true) -> "foo bar" | |
101 // * CombineAndCollapseWhitespace("foo ", "bar", false) -> "foo bar" | |
102 // * CombineAndCollapseWhitespace("foo", " bar", false) -> "foo bar" | |
103 // * CombineAndCollapseWhitespace("foo", " bar", true) -> "foo bar" | |
104 // * CombineAndCollapseWhitespace("foo ", " bar", false) -> "foo bar" | |
105 // * CombineAndCollapseWhitespace(" foo", "bar ", false) -> " foobar " | |
106 // * CombineAndCollapseWhitespace(" foo", "bar ", true) -> " foo bar " | |
107 const base::string16 CombineAndCollapseWhitespace( | |
108 const base::string16& prefix, | |
109 const base::string16& suffix, | |
110 bool force_whitespace) { | |
111 base::string16 prefix_trimmed; | |
112 TrimPositions prefix_trailing_whitespace = | |
113 TrimWhitespace(prefix, TRIM_TRAILING, &prefix_trimmed); | |
114 | |
115 // Recursively compute the children's text. | |
116 base::string16 suffix_trimmed; | |
117 TrimPositions suffix_leading_whitespace = | |
118 TrimWhitespace(suffix, TRIM_LEADING, &suffix_trimmed); | |
119 | |
120 if (prefix_trailing_whitespace || suffix_leading_whitespace || | |
121 force_whitespace) { | |
122 return prefix_trimmed + ASCIIToUTF16(" ") + suffix_trimmed; | |
123 } else { | |
124 return prefix_trimmed + suffix_trimmed; | |
125 } | |
126 } | |
127 | |
128 // This is a helper function for the FindChildText() function (see below). | |
129 // Search depth is limited with the |depth| parameter. | |
130 base::string16 FindChildTextInner(const WebNode& node, int depth) { | |
131 if (depth <= 0 || node.isNull()) | |
132 return base::string16(); | |
133 | |
134 // Skip over comments. | |
135 if (node.nodeType() == WebNode::CommentNode) | |
136 return FindChildTextInner(node.nextSibling(), depth - 1); | |
137 | |
138 if (node.nodeType() != WebNode::ElementNode && | |
139 node.nodeType() != WebNode::TextNode) | |
140 return base::string16(); | |
141 | |
142 // Ignore elements known not to contain inferable labels. | |
143 if (node.isElementNode()) { | |
144 const WebElement element = node.toConst<WebElement>(); | |
145 if (IsOptionElement(element) || | |
146 IsScriptElement(element) || | |
147 IsNoScriptElement(element) || | |
148 (element.isFormControlElement() && | |
149 IsAutofillableElement(element.toConst<WebFormControlElement>()))) { | |
150 return base::string16(); | |
151 } | |
152 } | |
153 | |
154 // Extract the text exactly at this node. | |
155 base::string16 node_text = node.nodeValue(); | |
156 | |
157 // Recursively compute the children's text. | |
158 // Preserve inter-element whitespace separation. | |
159 base::string16 child_text = FindChildTextInner(node.firstChild(), depth - 1); | |
160 bool add_space = node.nodeType() == WebNode::TextNode && node_text.empty(); | |
161 node_text = CombineAndCollapseWhitespace(node_text, child_text, add_space); | |
162 | |
163 // Recursively compute the siblings' text. | |
164 // Again, preserve inter-element whitespace separation. | |
165 base::string16 sibling_text = | |
166 FindChildTextInner(node.nextSibling(), depth - 1); | |
167 add_space = node.nodeType() == WebNode::TextNode && node_text.empty(); | |
168 node_text = CombineAndCollapseWhitespace(node_text, sibling_text, add_space); | |
169 | |
170 return node_text; | |
171 } | |
172 | |
173 // Returns the aggregated values of the descendants of |element| that are | |
174 // non-empty text nodes. This is a faster alternative to |innerText()| for | |
175 // performance critical operations. It does a full depth-first search so can be | |
176 // used when the structure is not directly known. However, unlike with | |
177 // |innerText()|, the search depth and breadth are limited to a fixed threshold. | |
178 // Whitespace is trimmed from text accumulated at descendant nodes. | |
179 base::string16 FindChildText(const WebNode& node) { | |
180 if (node.isTextNode()) | |
181 return node.nodeValue(); | |
182 | |
183 WebNode child = node.firstChild(); | |
184 | |
185 const int kChildSearchDepth = 10; | |
186 base::string16 node_text = FindChildTextInner(child, kChildSearchDepth); | |
187 TrimWhitespace(node_text, TRIM_ALL, &node_text); | |
188 return node_text; | |
189 } | |
190 | |
191 // Helper for |InferLabelForElement()| that infers a label, if possible, from | |
192 // a previous sibling of |element|, | |
193 // e.g. Some Text <input ...> | |
194 // or Some <span>Text</span> <input ...> | |
195 // or <p>Some Text</p><input ...> | |
196 // or <label>Some Text</label> <input ...> | |
197 // or Some Text <img><input ...> | |
198 // or <b>Some Text</b><br/> <input ...>. | |
199 base::string16 InferLabelFromPrevious(const WebFormControlElement& element) { | |
200 base::string16 inferred_label; | |
201 WebNode previous = element; | |
202 while (true) { | |
203 previous = previous.previousSibling(); | |
204 if (previous.isNull()) | |
205 break; | |
206 | |
207 // Skip over comments. | |
208 WebNode::NodeType node_type = previous.nodeType(); | |
209 if (node_type == WebNode::CommentNode) | |
210 continue; | |
211 | |
212 // Otherwise, only consider normal HTML elements and their contents. | |
213 if (node_type != WebNode::TextNode && | |
214 node_type != WebNode::ElementNode) | |
215 break; | |
216 | |
217 // A label might be split across multiple "lightweight" nodes. | |
218 // Coalesce any text contained in multiple consecutive | |
219 // (a) plain text nodes or | |
220 // (b) inline HTML elements that are essentially equivalent to text nodes. | |
221 CR_DEFINE_STATIC_LOCAL(WebString, kBold, ("b")); | |
222 CR_DEFINE_STATIC_LOCAL(WebString, kStrong, ("strong")); | |
223 CR_DEFINE_STATIC_LOCAL(WebString, kSpan, ("span")); | |
224 CR_DEFINE_STATIC_LOCAL(WebString, kFont, ("font")); | |
225 if (previous.isTextNode() || | |
226 HasTagName(previous, kBold) || HasTagName(previous, kStrong) || | |
227 HasTagName(previous, kSpan) || HasTagName(previous, kFont)) { | |
228 base::string16 value = FindChildText(previous); | |
229 // A text node's value will be empty if it is for a line break. | |
230 bool add_space = previous.isTextNode() && value.empty(); | |
231 inferred_label = | |
232 CombineAndCollapseWhitespace(value, inferred_label, add_space); | |
233 continue; | |
234 } | |
235 | |
236 // If we have identified a partial label and have reached a non-lightweight | |
237 // element, consider the label to be complete. | |
238 base::string16 trimmed_label; | |
239 TrimWhitespace(inferred_label, TRIM_ALL, &trimmed_label); | |
240 if (!trimmed_label.empty()) | |
241 break; | |
242 | |
243 // <img> and <br> tags often appear between the input element and its | |
244 // label text, so skip over them. | |
245 CR_DEFINE_STATIC_LOCAL(WebString, kImage, ("img")); | |
246 CR_DEFINE_STATIC_LOCAL(WebString, kBreak, ("br")); | |
247 if (HasTagName(previous, kImage) || HasTagName(previous, kBreak)) | |
248 continue; | |
249 | |
250 // We only expect <p> and <label> tags to contain the full label text. | |
251 CR_DEFINE_STATIC_LOCAL(WebString, kPage, ("p")); | |
252 CR_DEFINE_STATIC_LOCAL(WebString, kLabel, ("label")); | |
253 if (HasTagName(previous, kPage) || HasTagName(previous, kLabel)) | |
254 inferred_label = FindChildText(previous); | |
255 | |
256 break; | |
257 } | |
258 | |
259 TrimWhitespace(inferred_label, TRIM_ALL, &inferred_label); | |
260 return inferred_label; | |
261 } | |
262 | |
263 // Helper for |InferLabelForElement()| that infers a label, if possible, from | |
264 // enclosing list item, | |
265 // e.g. <li>Some Text<input ...><input ...><input ...></tr> | |
266 base::string16 InferLabelFromListItem(const WebFormControlElement& element) { | |
267 WebNode parent = element.parentNode(); | |
268 CR_DEFINE_STATIC_LOCAL(WebString, kListItem, ("li")); | |
269 while (!parent.isNull() && parent.isElementNode() && | |
270 !parent.to<WebElement>().hasTagName(kListItem)) { | |
271 parent = parent.parentNode(); | |
272 } | |
273 | |
274 if (!parent.isNull() && HasTagName(parent, kListItem)) | |
275 return FindChildText(parent); | |
276 | |
277 return base::string16(); | |
278 } | |
279 | |
280 // Helper for |InferLabelForElement()| that infers a label, if possible, from | |
281 // surrounding table structure, | |
282 // e.g. <tr><td>Some Text</td><td><input ...></td></tr> | |
283 // or <tr><th>Some Text</th><td><input ...></td></tr> | |
284 // or <tr><td><b>Some Text</b></td><td><b><input ...></b></td></tr> | |
285 // or <tr><th><b>Some Text</b></th><td><b><input ...></b></td></tr> | |
286 base::string16 InferLabelFromTableColumn(const WebFormControlElement& element) { | |
287 CR_DEFINE_STATIC_LOCAL(WebString, kTableCell, ("td")); | |
288 WebNode parent = element.parentNode(); | |
289 while (!parent.isNull() && parent.isElementNode() && | |
290 !parent.to<WebElement>().hasTagName(kTableCell)) { | |
291 parent = parent.parentNode(); | |
292 } | |
293 | |
294 if (parent.isNull()) | |
295 return base::string16(); | |
296 | |
297 // Check all previous siblings, skipping non-element nodes, until we find a | |
298 // non-empty text block. | |
299 base::string16 inferred_label; | |
300 WebNode previous = parent.previousSibling(); | |
301 CR_DEFINE_STATIC_LOCAL(WebString, kTableHeader, ("th")); | |
302 while (inferred_label.empty() && !previous.isNull()) { | |
303 if (HasTagName(previous, kTableCell) || HasTagName(previous, kTableHeader)) | |
304 inferred_label = FindChildText(previous); | |
305 | |
306 previous = previous.previousSibling(); | |
307 } | |
308 | |
309 return inferred_label; | |
310 } | |
311 | |
312 // Helper for |InferLabelForElement()| that infers a label, if possible, from | |
313 // surrounding table structure, | |
314 // e.g. <tr><td>Some Text</td></tr><tr><td><input ...></td></tr> | |
315 base::string16 InferLabelFromTableRow(const WebFormControlElement& element) { | |
316 CR_DEFINE_STATIC_LOCAL(WebString, kTableRow, ("tr")); | |
317 WebNode parent = element.parentNode(); | |
318 while (!parent.isNull() && parent.isElementNode() && | |
319 !parent.to<WebElement>().hasTagName(kTableRow)) { | |
320 parent = parent.parentNode(); | |
321 } | |
322 | |
323 if (parent.isNull()) | |
324 return base::string16(); | |
325 | |
326 // Check all previous siblings, skipping non-element nodes, until we find a | |
327 // non-empty text block. | |
328 base::string16 inferred_label; | |
329 WebNode previous = parent.previousSibling(); | |
330 while (inferred_label.empty() && !previous.isNull()) { | |
331 if (HasTagName(previous, kTableRow)) | |
332 inferred_label = FindChildText(previous); | |
333 | |
334 previous = previous.previousSibling(); | |
335 } | |
336 | |
337 return inferred_label; | |
338 } | |
339 | |
340 // Helper for |InferLabelForElement()| that infers a label, if possible, from | |
341 // a surrounding div table, | |
342 // e.g. <div>Some Text<span><input ...></span></div> | |
343 // e.g. <div>Some Text</div><div><input ...></div> | |
344 base::string16 InferLabelFromDivTable(const WebFormControlElement& element) { | |
345 WebNode node = element.parentNode(); | |
346 bool looking_for_parent = true; | |
347 | |
348 // Search the sibling and parent <div>s until we find a candidate label. | |
349 base::string16 inferred_label; | |
350 CR_DEFINE_STATIC_LOCAL(WebString, kDiv, ("div")); | |
351 CR_DEFINE_STATIC_LOCAL(WebString, kTable, ("table")); | |
352 CR_DEFINE_STATIC_LOCAL(WebString, kFieldSet, ("fieldset")); | |
353 while (inferred_label.empty() && !node.isNull()) { | |
354 if (HasTagName(node, kDiv)) { | |
355 looking_for_parent = false; | |
356 inferred_label = FindChildText(node); | |
357 } else if (looking_for_parent && | |
358 (HasTagName(node, kTable) || HasTagName(node, kFieldSet))) { | |
359 // If the element is in a table or fieldset, its label most likely is too. | |
360 break; | |
361 } | |
362 | |
363 if (node.previousSibling().isNull()) { | |
364 // If there are no more siblings, continue walking up the tree. | |
365 looking_for_parent = true; | |
366 } | |
367 | |
368 if (looking_for_parent) | |
369 node = node.parentNode(); | |
370 else | |
371 node = node.previousSibling(); | |
372 } | |
373 | |
374 return inferred_label; | |
375 } | |
376 | |
377 // Helper for |InferLabelForElement()| that infers a label, if possible, from | |
378 // a surrounding definition list, | |
379 // e.g. <dl><dt>Some Text</dt><dd><input ...></dd></dl> | |
380 // e.g. <dl><dt><b>Some Text</b></dt><dd><b><input ...></b></dd></dl> | |
381 base::string16 InferLabelFromDefinitionList( | |
382 const WebFormControlElement& element) { | |
383 CR_DEFINE_STATIC_LOCAL(WebString, kDefinitionData, ("dd")); | |
384 WebNode parent = element.parentNode(); | |
385 while (!parent.isNull() && parent.isElementNode() && | |
386 !parent.to<WebElement>().hasTagName(kDefinitionData)) | |
387 parent = parent.parentNode(); | |
388 | |
389 if (parent.isNull() || !HasTagName(parent, kDefinitionData)) | |
390 return base::string16(); | |
391 | |
392 // Skip by any intervening text nodes. | |
393 WebNode previous = parent.previousSibling(); | |
394 while (!previous.isNull() && previous.isTextNode()) | |
395 previous = previous.previousSibling(); | |
396 | |
397 CR_DEFINE_STATIC_LOCAL(WebString, kDefinitionTag, ("dt")); | |
398 if (previous.isNull() || !HasTagName(previous, kDefinitionTag)) | |
399 return base::string16(); | |
400 | |
401 return FindChildText(previous); | |
402 } | |
403 | |
404 // Infers corresponding label for |element| from surrounding context in the DOM, | |
405 // e.g. the contents of the preceding <p> tag or text element. | |
406 base::string16 InferLabelForElement(const WebFormControlElement& element) { | |
407 base::string16 inferred_label = InferLabelFromPrevious(element); | |
408 if (!inferred_label.empty()) | |
409 return inferred_label; | |
410 | |
411 // If we didn't find a label, check for list item case. | |
412 inferred_label = InferLabelFromListItem(element); | |
413 if (!inferred_label.empty()) | |
414 return inferred_label; | |
415 | |
416 // If we didn't find a label, check for table cell case. | |
417 inferred_label = InferLabelFromTableColumn(element); | |
418 if (!inferred_label.empty()) | |
419 return inferred_label; | |
420 | |
421 // If we didn't find a label, check for table row case. | |
422 inferred_label = InferLabelFromTableRow(element); | |
423 if (!inferred_label.empty()) | |
424 return inferred_label; | |
425 | |
426 // If we didn't find a label, check for definition list case. | |
427 inferred_label = InferLabelFromDefinitionList(element); | |
428 if (!inferred_label.empty()) | |
429 return inferred_label; | |
430 | |
431 // If we didn't find a label, check for div table case. | |
432 return InferLabelFromDivTable(element); | |
433 } | |
434 | |
435 // Fills |option_strings| with the values of the <option> elements present in | |
436 // |select_element|. | |
437 void GetOptionStringsFromElement(const WebSelectElement& select_element, | |
438 std::vector<base::string16>* option_values, | |
439 std::vector<base::string16>* option_contents) { | |
440 DCHECK(!select_element.isNull()); | |
441 | |
442 option_values->clear(); | |
443 option_contents->clear(); | |
444 WebVector<WebElement> list_items = select_element.listItems(); | |
445 option_values->reserve(list_items.size()); | |
446 option_contents->reserve(list_items.size()); | |
447 for (size_t i = 0; i < list_items.size(); ++i) { | |
448 if (IsOptionElement(list_items[i])) { | |
449 const WebOptionElement option = list_items[i].toConst<WebOptionElement>(); | |
450 option_values->push_back(option.value()); | |
451 option_contents->push_back(option.text()); | |
452 } | |
453 } | |
454 } | |
455 | |
456 // The callback type used by |ForEachMatchingFormField()|. | |
457 typedef void (*Callback)(const FormFieldData&, | |
458 bool, /* is_initiating_element */ | |
459 WebKit::WebFormControlElement*); | |
460 | |
461 // For each autofillable field in |data| that matches a field in the |form|, | |
462 // the |callback| is invoked with the corresponding |form| field data. | |
463 void ForEachMatchingFormField(const WebFormElement& form_element, | |
464 const WebElement& initiating_element, | |
465 const FormData& data, | |
466 bool only_focusable_elements, | |
467 bool force_override, | |
468 Callback callback) { | |
469 std::vector<WebFormControlElement> control_elements; | |
470 ExtractAutofillableElements(form_element, REQUIRE_AUTOCOMPLETE, | |
471 &control_elements); | |
472 | |
473 if (control_elements.size() != data.fields.size()) { | |
474 // This case should be reachable only for pathological websites and tests, | |
475 // which add or remove form fields while the user is interacting with the | |
476 // Autofill popup. | |
477 return; | |
478 } | |
479 | |
480 // It's possible that the site has injected fields into the form after the | |
481 // page has loaded, so we can't assert that the size of the cached control | |
482 // elements is equal to the size of the fields in |form|. Fortunately, the | |
483 // one case in the wild where this happens, paypal.com signup form, the fields | |
484 // are appended to the end of the form and are not visible. | |
485 for (size_t i = 0; i < control_elements.size(); ++i) { | |
486 WebFormControlElement* element = &control_elements[i]; | |
487 | |
488 if (base::string16(element->nameForAutofill()) != data.fields[i].name) { | |
489 // This case should be reachable only for pathological websites, which | |
490 // rename form fields while the user is interacting with the Autofill | |
491 // popup. I (isherman) am not aware of any such websites, and so am | |
492 // optimistically including a NOTREACHED(). If you ever trip this check, | |
493 // please file a bug against me. | |
494 NOTREACHED(); | |
495 continue; | |
496 } | |
497 | |
498 bool is_initiating_element = (*element == initiating_element); | |
499 | |
500 // Only autofill empty fields and the field that initiated the filling, | |
501 // i.e. the field the user is currently editing and interacting with. | |
502 const WebInputElement* input_element = toWebInputElement(element); | |
503 if (!force_override && IsTextInput(input_element) && | |
504 !is_initiating_element && !input_element->value().isEmpty()) | |
505 continue; | |
506 | |
507 if (!element->isEnabled() || element->isReadOnly() || | |
508 (only_focusable_elements && !element->isFocusable())) | |
509 continue; | |
510 | |
511 callback(data.fields[i], is_initiating_element, element); | |
512 } | |
513 } | |
514 | |
515 // Sets the |field|'s value to the value in |data|. | |
516 // Also sets the "autofilled" attribute, causing the background to be yellow. | |
517 void FillFormField(const FormFieldData& data, | |
518 bool is_initiating_node, | |
519 WebKit::WebFormControlElement* field) { | |
520 // Nothing to fill. | |
521 if (data.value.empty()) | |
522 return; | |
523 | |
524 WebInputElement* input_element = toWebInputElement(field); | |
525 if (IsTextInput(input_element)) { | |
526 // If the maxlength attribute contains a negative value, maxLength() | |
527 // returns the default maxlength value. | |
528 input_element->setValue( | |
529 data.value.substr(0, input_element->maxLength()), true); | |
530 input_element->setAutofilled(true); | |
531 if (is_initiating_node) { | |
532 int length = input_element->value().length(); | |
533 input_element->setSelectionRange(length, length); | |
534 // Clear the current IME composition (the underline), if there is one. | |
535 input_element->document().frame()->unmarkText(); | |
536 } | |
537 } else if (IsSelectElement(*field)) { | |
538 WebSelectElement select_element = field->to<WebSelectElement>(); | |
539 if (select_element.value() != data.value) { | |
540 select_element.setValue(data.value); | |
541 select_element.dispatchFormControlChangeEvent(); | |
542 } | |
543 } else { | |
544 DCHECK(IsCheckableElement(input_element)); | |
545 input_element->setChecked(data.is_checked, true); | |
546 } | |
547 } | |
548 | |
549 // Sets the |field|'s "suggested" (non JS visible) value to the value in |data|. | |
550 // Also sets the "autofilled" attribute, causing the background to be yellow. | |
551 void PreviewFormField(const FormFieldData& data, | |
552 bool is_initiating_node, | |
553 WebKit::WebFormControlElement* field) { | |
554 // Nothing to preview. | |
555 if (data.value.empty()) | |
556 return; | |
557 | |
558 // Only preview input fields. Excludes checkboxes and radio buttons, as there | |
559 // is no provision for setSuggestedCheckedValue in WebInputElement. | |
560 WebInputElement* input_element = toWebInputElement(field); | |
561 if (!IsTextInput(input_element)) | |
562 return; | |
563 | |
564 // If the maxlength attribute contains a negative value, maxLength() | |
565 // returns the default maxlength value. | |
566 input_element->setSuggestedValue( | |
567 data.value.substr(0, input_element->maxLength())); | |
568 input_element->setAutofilled(true); | |
569 if (is_initiating_node) { | |
570 // Select the part of the text that the user didn't type. | |
571 input_element->setSelectionRange(input_element->value().length(), | |
572 input_element->suggestedValue().length()); | |
573 } | |
574 } | |
575 | |
576 std::string RetrievalMethodToString( | |
577 const WebElementDescriptor::RetrievalMethod& method) { | |
578 switch (method) { | |
579 case WebElementDescriptor::CSS_SELECTOR: | |
580 return "CSS_SELECTOR"; | |
581 case WebElementDescriptor::ID: | |
582 return "ID"; | |
583 case WebElementDescriptor::NONE: | |
584 return "NONE"; | |
585 } | |
586 NOTREACHED(); | |
587 return "UNKNOWN"; | |
588 } | |
589 | |
590 } // namespace | |
591 | |
592 const size_t kMaxParseableFields = 100; | |
593 | |
594 // All text fields, including password fields, should be extracted. | |
595 bool IsTextInput(const WebInputElement* element) { | |
596 return element && element->isTextField(); | |
597 } | |
598 | |
599 bool IsSelectElement(const WebFormControlElement& element) { | |
600 // Is static for improving performance. | |
601 CR_DEFINE_STATIC_LOCAL(WebString, kSelectOne, ("select-one")); | |
602 return element.formControlType() == kSelectOne; | |
603 } | |
604 | |
605 bool IsCheckableElement(const WebInputElement* element) { | |
606 if (!element) | |
607 return false; | |
608 | |
609 return element->isCheckbox() || element->isRadioButton(); | |
610 } | |
611 | |
612 bool IsAutofillableInputElement(const WebInputElement* element) { | |
613 return IsTextInput(element) || IsCheckableElement(element); | |
614 } | |
615 | |
616 const base::string16 GetFormIdentifier(const WebFormElement& form) { | |
617 base::string16 identifier = form.name(); | |
618 CR_DEFINE_STATIC_LOCAL(WebString, kId, ("id")); | |
619 if (identifier.empty()) | |
620 identifier = form.getAttribute(kId); | |
621 | |
622 return identifier; | |
623 } | |
624 | |
625 bool ClickElement(const WebDocument& document, | |
626 const WebElementDescriptor& element_descriptor) { | |
627 WebString web_descriptor = WebString::fromUTF8(element_descriptor.descriptor); | |
628 WebKit::WebElement element; | |
629 | |
630 switch (element_descriptor.retrieval_method) { | |
631 case WebElementDescriptor::CSS_SELECTOR: { | |
632 WebExceptionCode ec = 0; | |
633 element = document.querySelector(web_descriptor, ec); | |
634 if (ec) | |
635 DVLOG(1) << "Query selector failed. Error code: " << ec << "."; | |
636 break; | |
637 } | |
638 case WebElementDescriptor::ID: | |
639 element = document.getElementById(web_descriptor); | |
640 break; | |
641 case WebElementDescriptor::NONE: | |
642 return true; | |
643 } | |
644 | |
645 if (element.isNull()) { | |
646 DVLOG(1) << "Could not find " | |
647 << element_descriptor.descriptor | |
648 << " by " | |
649 << RetrievalMethodToString(element_descriptor.retrieval_method) | |
650 << "."; | |
651 return false; | |
652 } | |
653 | |
654 element.simulateClick(); | |
655 return true; | |
656 } | |
657 | |
658 // Fills |autofillable_elements| with all the auto-fillable form control | |
659 // elements in |form_element|. | |
660 void ExtractAutofillableElements( | |
661 const WebFormElement& form_element, | |
662 RequirementsMask requirements, | |
663 std::vector<WebFormControlElement>* autofillable_elements) { | |
664 WebVector<WebFormControlElement> control_elements; | |
665 form_element.getFormControlElements(control_elements); | |
666 | |
667 autofillable_elements->clear(); | |
668 for (size_t i = 0; i < control_elements.size(); ++i) { | |
669 WebFormControlElement element = control_elements[i]; | |
670 if (!IsAutofillableElement(element)) | |
671 continue; | |
672 | |
673 if (requirements & REQUIRE_AUTOCOMPLETE) { | |
674 // TODO(jhawkins): WebKit currently doesn't handle the autocomplete | |
675 // attribute for select control elements, but it probably should. | |
676 WebInputElement* input_element = toWebInputElement(&control_elements[i]); | |
677 if (IsAutofillableInputElement(input_element) && | |
678 !SatisfiesRequireAutocomplete(*input_element)) | |
679 continue; | |
680 } | |
681 | |
682 autofillable_elements->push_back(element); | |
683 } | |
684 } | |
685 | |
686 void WebFormControlElementToFormField(const WebFormControlElement& element, | |
687 ExtractMask extract_mask, | |
688 FormFieldData* field) { | |
689 DCHECK(field); | |
690 DCHECK(!element.isNull()); | |
691 CR_DEFINE_STATIC_LOCAL(WebString, kAutocomplete, ("autocomplete")); | |
692 | |
693 // The label is not officially part of a WebFormControlElement; however, the | |
694 // labels for all form control elements are scraped from the DOM and set in | |
695 // WebFormElementToFormData. | |
696 field->name = element.nameForAutofill(); | |
697 field->form_control_type = UTF16ToUTF8(element.formControlType()); | |
698 field->autocomplete_attribute = | |
699 UTF16ToUTF8(element.getAttribute(kAutocomplete)); | |
700 if (field->autocomplete_attribute.size() > kMaxDataLength) { | |
701 // Discard overly long attribute values to avoid DOS-ing the browser | |
702 // process. However, send over a default string to indicate that the | |
703 // attribute was present. | |
704 field->autocomplete_attribute = "x-max-data-length-exceeded"; | |
705 } | |
706 | |
707 if (!IsAutofillableElement(element)) | |
708 return; | |
709 | |
710 const WebInputElement* input_element = toWebInputElement(&element); | |
711 if (IsAutofillableInputElement(input_element)) { | |
712 if (IsTextInput(input_element)) | |
713 field->max_length = input_element->maxLength(); | |
714 | |
715 field->is_autofilled = input_element->isAutofilled(); | |
716 field->is_focusable = input_element->isFocusable(); | |
717 field->should_autocomplete = input_element->autoComplete(); | |
718 field->is_checkable = IsCheckableElement(input_element); | |
719 field->is_checked = input_element->isChecked(); | |
720 } else if (extract_mask & EXTRACT_OPTIONS) { | |
721 // Set option strings on the field if available. | |
722 DCHECK(IsSelectElement(element)); | |
723 const WebSelectElement select_element = element.toConst<WebSelectElement>(); | |
724 GetOptionStringsFromElement(select_element, | |
725 &field->option_values, | |
726 &field->option_contents); | |
727 } | |
728 | |
729 if (!(extract_mask & EXTRACT_VALUE)) | |
730 return; | |
731 | |
732 base::string16 value; | |
733 if (IsAutofillableInputElement(input_element)) { | |
734 value = input_element->value(); | |
735 } else { | |
736 DCHECK(IsSelectElement(element)); | |
737 const WebSelectElement select_element = element.toConst<WebSelectElement>(); | |
738 value = select_element.value(); | |
739 | |
740 // Convert the |select_element| value to text if requested. | |
741 if (extract_mask & EXTRACT_OPTION_TEXT) { | |
742 WebVector<WebElement> list_items = select_element.listItems(); | |
743 for (size_t i = 0; i < list_items.size(); ++i) { | |
744 if (IsOptionElement(list_items[i])) { | |
745 const WebOptionElement option_element = | |
746 list_items[i].toConst<WebOptionElement>(); | |
747 if (option_element.value() == value) { | |
748 value = option_element.text(); | |
749 break; | |
750 } | |
751 } | |
752 } | |
753 } | |
754 } | |
755 | |
756 // Constrain the maximum data length to prevent a malicious site from DOS'ing | |
757 // the browser: http://crbug.com/49332 | |
758 if (value.size() > kMaxDataLength) | |
759 value = value.substr(0, kMaxDataLength); | |
760 | |
761 field->value = value; | |
762 } | |
763 | |
764 bool WebFormElementToFormData( | |
765 const WebKit::WebFormElement& form_element, | |
766 const WebKit::WebFormControlElement& form_control_element, | |
767 RequirementsMask requirements, | |
768 ExtractMask extract_mask, | |
769 FormData* form, | |
770 FormFieldData* field) { | |
771 CR_DEFINE_STATIC_LOCAL(WebString, kLabel, ("label")); | |
772 CR_DEFINE_STATIC_LOCAL(WebString, kFor, ("for")); | |
773 CR_DEFINE_STATIC_LOCAL(WebString, kHidden, ("hidden")); | |
774 | |
775 const WebFrame* frame = form_element.document().frame(); | |
776 if (!frame) | |
777 return false; | |
778 | |
779 if (requirements & REQUIRE_AUTOCOMPLETE && !form_element.autoComplete()) | |
780 return false; | |
781 | |
782 form->name = GetFormIdentifier(form_element); | |
783 form->method = form_element.method(); | |
784 form->origin = frame->document().url(); | |
785 form->action = frame->document().completeURL(form_element.action()); | |
786 form->user_submitted = form_element.wasUserSubmitted(); | |
787 | |
788 // If the completed URL is not valid, just use the action we get from | |
789 // WebKit. | |
790 if (!form->action.is_valid()) | |
791 form->action = GURL(form_element.action()); | |
792 | |
793 // A map from a FormFieldData's name to the FormFieldData itself. | |
794 std::map<base::string16, FormFieldData*> name_map; | |
795 | |
796 // The extracted FormFields. We use pointers so we can store them in | |
797 // |name_map|. | |
798 ScopedVector<FormFieldData> form_fields; | |
799 | |
800 WebVector<WebFormControlElement> control_elements; | |
801 form_element.getFormControlElements(control_elements); | |
802 | |
803 // A vector of bools that indicate whether each field in the form meets the | |
804 // requirements and thus will be in the resulting |form|. | |
805 std::vector<bool> fields_extracted(control_elements.size(), false); | |
806 | |
807 for (size_t i = 0; i < control_elements.size(); ++i) { | |
808 const WebFormControlElement& control_element = control_elements[i]; | |
809 | |
810 if (!IsAutofillableElement(control_element)) | |
811 continue; | |
812 | |
813 const WebInputElement* input_element = toWebInputElement(&control_element); | |
814 if (requirements & REQUIRE_AUTOCOMPLETE && | |
815 IsAutofillableInputElement(input_element) && | |
816 !SatisfiesRequireAutocomplete(*input_element)) | |
817 continue; | |
818 | |
819 // Create a new FormFieldData, fill it out and map it to the field's name. | |
820 FormFieldData* form_field = new FormFieldData; | |
821 WebFormControlElementToFormField(control_element, extract_mask, form_field); | |
822 form_fields.push_back(form_field); | |
823 // TODO(jhawkins): A label element is mapped to a form control element's id. | |
824 // field->name() will contain the id only if the name does not exist. Add | |
825 // an id() method to WebFormControlElement and use that here. | |
826 name_map[form_field->name] = form_field; | |
827 fields_extracted[i] = true; | |
828 } | |
829 | |
830 // If we failed to extract any fields, give up. Also, to avoid overly | |
831 // expensive computation, we impose a maximum number of allowable fields. | |
832 if (form_fields.empty() || form_fields.size() > kMaxParseableFields) | |
833 return false; | |
834 | |
835 // Loop through the label elements inside the form element. For each label | |
836 // element, get the corresponding form control element, use the form control | |
837 // element's name as a key into the <name, FormFieldData> map to find the | |
838 // previously created FormFieldData and set the FormFieldData's label to the | |
839 // label.firstChild().nodeValue() of the label element. | |
840 WebNodeList labels = form_element.getElementsByTagName(kLabel); | |
841 for (unsigned i = 0; i < labels.length(); ++i) { | |
842 WebLabelElement label = labels.item(i).to<WebLabelElement>(); | |
843 WebFormControlElement field_element = | |
844 label.correspondingControl().to<WebFormControlElement>(); | |
845 | |
846 base::string16 element_name; | |
847 if (field_element.isNull()) { | |
848 // Sometimes site authors will incorrectly specify the corresponding | |
849 // field element's name rather than its id, so we compensate here. | |
850 element_name = label.getAttribute(kFor); | |
851 } else if ( | |
852 !field_element.isFormControlElement() || | |
853 field_element.formControlType() == kHidden) { | |
854 continue; | |
855 } else { | |
856 element_name = field_element.nameForAutofill(); | |
857 } | |
858 | |
859 std::map<base::string16, FormFieldData*>::iterator iter = | |
860 name_map.find(element_name); | |
861 if (iter != name_map.end()) { | |
862 base::string16 label_text = FindChildText(label); | |
863 | |
864 // Concatenate labels because some sites might have multiple label | |
865 // candidates. | |
866 if (!iter->second->label.empty() && !label_text.empty()) | |
867 iter->second->label += ASCIIToUTF16(" "); | |
868 iter->second->label += label_text; | |
869 } | |
870 } | |
871 | |
872 // Loop through the form control elements, extracting the label text from | |
873 // the DOM. We use the |fields_extracted| vector to make sure we assign the | |
874 // extracted label to the correct field, as it's possible |form_fields| will | |
875 // not contain all of the elements in |control_elements|. | |
876 for (size_t i = 0, field_idx = 0; | |
877 i < control_elements.size() && field_idx < form_fields.size(); ++i) { | |
878 // This field didn't meet the requirements, so don't try to find a label | |
879 // for it. | |
880 if (!fields_extracted[i]) | |
881 continue; | |
882 | |
883 const WebFormControlElement& control_element = control_elements[i]; | |
884 if (form_fields[field_idx]->label.empty()) | |
885 form_fields[field_idx]->label = InferLabelForElement(control_element); | |
886 | |
887 if (field && form_control_element == control_element) | |
888 *field = *form_fields[field_idx]; | |
889 | |
890 ++field_idx; | |
891 } | |
892 | |
893 // Copy the created FormFields into the resulting FormData object. | |
894 for (ScopedVector<FormFieldData>::const_iterator iter = form_fields.begin(); | |
895 iter != form_fields.end(); ++iter) { | |
896 form->fields.push_back(**iter); | |
897 } | |
898 | |
899 return true; | |
900 } | |
901 | |
902 bool FindFormAndFieldForInputElement(const WebInputElement& element, | |
903 FormData* form, | |
904 FormFieldData* field, | |
905 RequirementsMask requirements) { | |
906 if (!IsAutofillableElement(element)) | |
907 return false; | |
908 | |
909 const WebFormElement form_element = element.form(); | |
910 if (form_element.isNull()) | |
911 return false; | |
912 | |
913 ExtractMask extract_mask = | |
914 static_cast<ExtractMask>(EXTRACT_VALUE | EXTRACT_OPTIONS); | |
915 return WebFormElementToFormData(form_element, | |
916 element, | |
917 requirements, | |
918 extract_mask, | |
919 form, | |
920 field); | |
921 } | |
922 | |
923 void FillForm(const FormData& form, const WebInputElement& element) { | |
924 WebFormElement form_element = element.form(); | |
925 if (form_element.isNull()) | |
926 return; | |
927 | |
928 ForEachMatchingFormField(form_element, | |
929 element, | |
930 form, | |
931 true, /* only_focusable_elements */ | |
932 false, /* dont force override */ | |
933 &FillFormField); | |
934 } | |
935 | |
936 void FillFormIncludingNonFocusableElements(const FormData& form_data, | |
937 const WebFormElement& form_element) { | |
938 if (form_element.isNull()) | |
939 return; | |
940 | |
941 ForEachMatchingFormField(form_element, | |
942 WebInputElement(), | |
943 form_data, | |
944 false, /* only_focusable_elements */ | |
945 true, /* force override */ | |
946 &FillFormField); | |
947 } | |
948 | |
949 void PreviewForm(const FormData& form, const WebInputElement& element) { | |
950 WebFormElement form_element = element.form(); | |
951 if (form_element.isNull()) | |
952 return; | |
953 | |
954 ForEachMatchingFormField(form_element, | |
955 element, | |
956 form, | |
957 true, /* only_focusable_elements */ | |
958 false, /* dont force override */ | |
959 &PreviewFormField); | |
960 } | |
961 | |
962 bool ClearPreviewedFormWithElement(const WebInputElement& element, | |
963 bool was_autofilled) { | |
964 WebFormElement form_element = element.form(); | |
965 if (form_element.isNull()) | |
966 return false; | |
967 | |
968 std::vector<WebFormControlElement> control_elements; | |
969 ExtractAutofillableElements(form_element, REQUIRE_AUTOCOMPLETE, | |
970 &control_elements); | |
971 for (size_t i = 0; i < control_elements.size(); ++i) { | |
972 // Only text input elements can be previewed. | |
973 WebInputElement* input_element = toWebInputElement(&control_elements[i]); | |
974 if (!IsTextInput(input_element)) | |
975 continue; | |
976 | |
977 // If the input element is not auto-filled, we did not preview it, so there | |
978 // is nothing to reset. | |
979 if (!input_element->isAutofilled()) | |
980 continue; | |
981 | |
982 // There might be unrelated elements in this form which have already been | |
983 // auto-filled. For example, the user might have already filled the address | |
984 // part of a form and now be dealing with the credit card section. We only | |
985 // want to reset the auto-filled status for fields that were previewed. | |
986 if (input_element->suggestedValue().isEmpty()) | |
987 continue; | |
988 | |
989 // Clear the suggested value. For the initiating node, also restore the | |
990 // original value. | |
991 input_element->setSuggestedValue(WebString()); | |
992 bool is_initiating_node = (element == *input_element); | |
993 if (is_initiating_node) | |
994 input_element->setAutofilled(was_autofilled); | |
995 else | |
996 input_element->setAutofilled(false); | |
997 | |
998 // Clearing the suggested value in the focused node (above) can cause | |
999 // selection to be lost. We force selection range to restore the text | |
1000 // cursor. | |
1001 if (is_initiating_node) { | |
1002 int length = input_element->value().length(); | |
1003 input_element->setSelectionRange(length, length); | |
1004 } | |
1005 } | |
1006 | |
1007 return true; | |
1008 } | |
1009 | |
1010 bool FormWithElementIsAutofilled(const WebInputElement& element) { | |
1011 WebFormElement form_element = element.form(); | |
1012 if (form_element.isNull()) | |
1013 return false; | |
1014 | |
1015 std::vector<WebFormControlElement> control_elements; | |
1016 ExtractAutofillableElements(form_element, REQUIRE_AUTOCOMPLETE, | |
1017 &control_elements); | |
1018 for (size_t i = 0; i < control_elements.size(); ++i) { | |
1019 WebInputElement* input_element = toWebInputElement(&control_elements[i]); | |
1020 if (!IsAutofillableInputElement(input_element)) | |
1021 continue; | |
1022 | |
1023 if (input_element->isAutofilled()) | |
1024 return true; | |
1025 } | |
1026 | |
1027 return false; | |
1028 } | |
1029 | |
1030 } // namespace autofill | |
OLD | NEW |