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 "chrome/renderer/autofill/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/string_util.h" | |
13 #include "base/utf_string_conversions.h" | |
14 #include "components/autofill/common/autofill_switches.h" | |
15 #include "components/autofill/common/form_data.h" | |
16 #include "components/autofill/common/form_field_data.h" | |
17 #include "components/autofill/common/web_element_descriptor.h" | |
18 #include "third_party/WebKit/Source/Platform/chromium/public/WebString.h" | |
19 #include "third_party/WebKit/Source/Platform/chromium/public/WebVector.h" | |
20 #include "third_party/WebKit/Source/WebKit/chromium/public/WebDocument.h" | |
21 #include "third_party/WebKit/Source/WebKit/chromium/public/WebElement.h" | |
22 #include "third_party/WebKit/Source/WebKit/chromium/public/WebExceptionCode.h" | |
23 #include "third_party/WebKit/Source/WebKit/chromium/public/WebFormControlElement
.h" | |
24 #include "third_party/WebKit/Source/WebKit/chromium/public/WebFormElement.h" | |
25 #include "third_party/WebKit/Source/WebKit/chromium/public/WebFrame.h" | |
26 #include "third_party/WebKit/Source/WebKit/chromium/public/WebInputElement.h" | |
27 #include "third_party/WebKit/Source/WebKit/chromium/public/WebLabelElement.h" | |
28 #include "third_party/WebKit/Source/WebKit/chromium/public/WebNode.h" | |
29 #include "third_party/WebKit/Source/WebKit/chromium/public/WebNodeList.h" | |
30 #include "third_party/WebKit/Source/WebKit/chromium/public/WebOptionElement.h" | |
31 #include "third_party/WebKit/Source/WebKit/chromium/public/WebSelectElement.h" | |
32 | |
33 using WebKit::WebDocument; | |
34 using WebKit::WebElement; | |
35 using WebKit::WebExceptionCode; | |
36 using WebKit::WebFormControlElement; | |
37 using WebKit::WebFormElement; | |
38 using WebKit::WebFrame; | |
39 using WebKit::WebInputElement; | |
40 using WebKit::WebLabelElement; | |
41 using WebKit::WebNode; | |
42 using WebKit::WebNodeList; | |
43 using WebKit::WebOptionElement; | |
44 using WebKit::WebSelectElement; | |
45 using WebKit::WebString; | |
46 using WebKit::WebVector; | |
47 | |
48 namespace { | |
49 | |
50 using autofill::ExtractAutofillableElements; | |
51 using autofill::IsAutofillableInputElement; | |
52 using autofill::IsCheckableElement; | |
53 using autofill::IsSelectElement; | |
54 using autofill::IsTextInput; | |
55 | |
56 // The maximum length allowed for form data. | |
57 const size_t kMaxDataLength = 1024; | |
58 | |
59 bool IsOptionElement(const WebElement& element) { | |
60 CR_DEFINE_STATIC_LOCAL(WebString, kOption, ("option")); | |
61 return element.hasTagName(kOption); | |
62 } | |
63 | |
64 bool IsScriptElement(const WebElement& element) { | |
65 CR_DEFINE_STATIC_LOCAL(WebString, kScript, ("script")); | |
66 return element.hasTagName(kScript); | |
67 } | |
68 | |
69 bool IsNoScriptElement(const WebElement& element) { | |
70 CR_DEFINE_STATIC_LOCAL(WebString, kNoScript, ("noscript")); | |
71 return element.hasTagName(kNoScript); | |
72 } | |
73 | |
74 bool HasTagName(const WebNode& node, const WebKit::WebString& tag) { | |
75 return node.isElementNode() && node.toConst<WebElement>().hasTagName(tag); | |
76 } | |
77 | |
78 bool IsAutofillableElement(const WebFormControlElement& element) { | |
79 const WebInputElement* input_element = toWebInputElement(&element); | |
80 return IsAutofillableInputElement(input_element) || IsSelectElement(element); | |
81 } | |
82 | |
83 // Check whether the given field satisfies the REQUIRE_AUTOCOMPLETE requirement. | |
84 // When Autocheckout is enabled, this requirement is enforced in the browser | |
85 // process rather than in the renderer process, and hence all fields are | |
86 // considered to satisfy this requirement. | |
87 bool SatisfiesRequireAutocomplete(const WebInputElement& input_element) { | |
88 return input_element.autoComplete() || | |
89 CommandLine::ForCurrentProcess()->HasSwitch( | |
90 switches::kEnableExperimentalFormFilling); | |
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 string16 CombineAndCollapseWhitespace(const string16& prefix, | |
108 const string16& suffix, | |
109 bool force_whitespace) { | |
110 string16 prefix_trimmed; | |
111 TrimPositions prefix_trailing_whitespace = | |
112 TrimWhitespace(prefix, TRIM_TRAILING, &prefix_trimmed); | |
113 | |
114 // Recursively compute the children's text. | |
115 string16 suffix_trimmed; | |
116 TrimPositions suffix_leading_whitespace = | |
117 TrimWhitespace(suffix, TRIM_LEADING, &suffix_trimmed); | |
118 | |
119 if (prefix_trailing_whitespace || suffix_leading_whitespace || | |
120 force_whitespace) { | |
121 return prefix_trimmed + ASCIIToUTF16(" ") + suffix_trimmed; | |
122 } else { | |
123 return prefix_trimmed + suffix_trimmed; | |
124 } | |
125 } | |
126 | |
127 // This is a helper function for the FindChildText() function (see below). | |
128 // Search depth is limited with the |depth| parameter. | |
129 string16 FindChildTextInner(const WebNode& node, int depth) { | |
130 if (depth <= 0 || node.isNull()) | |
131 return string16(); | |
132 | |
133 // Skip over comments. | |
134 if (node.nodeType() == WebNode::CommentNode) | |
135 return FindChildTextInner(node.nextSibling(), depth - 1); | |
136 | |
137 if (node.nodeType() != WebNode::ElementNode && | |
138 node.nodeType() != WebNode::TextNode) | |
139 return string16(); | |
140 | |
141 // Ignore elements known not to contain inferable labels. | |
142 if (node.isElementNode()) { | |
143 const WebElement element = node.toConst<WebElement>(); | |
144 if (IsOptionElement(element) || | |
145 IsScriptElement(element) || | |
146 IsNoScriptElement(element) || | |
147 (element.isFormControlElement() && | |
148 IsAutofillableElement(element.toConst<WebFormControlElement>()))) { | |
149 return string16(); | |
150 } | |
151 } | |
152 | |
153 // Extract the text exactly at this node. | |
154 string16 node_text = node.nodeValue(); | |
155 | |
156 // Recursively compute the children's text. | |
157 // Preserve inter-element whitespace separation. | |
158 string16 child_text = FindChildTextInner(node.firstChild(), depth - 1); | |
159 bool add_space = node.nodeType() == WebNode::TextNode && node_text.empty(); | |
160 node_text = CombineAndCollapseWhitespace(node_text, child_text, add_space); | |
161 | |
162 // Recursively compute the siblings' text. | |
163 // Again, preserve inter-element whitespace separation. | |
164 string16 sibling_text = FindChildTextInner(node.nextSibling(), depth - 1); | |
165 add_space = node.nodeType() == WebNode::TextNode && node_text.empty(); | |
166 node_text = CombineAndCollapseWhitespace(node_text, sibling_text, add_space); | |
167 | |
168 return node_text; | |
169 } | |
170 | |
171 // Returns the aggregated values of the descendants of |element| that are | |
172 // non-empty text nodes. This is a faster alternative to |innerText()| for | |
173 // performance critical operations. It does a full depth-first search so can be | |
174 // used when the structure is not directly known. However, unlike with | |
175 // |innerText()|, the search depth and breadth are limited to a fixed threshold. | |
176 // Whitespace is trimmed from text accumulated at descendant nodes. | |
177 string16 FindChildText(const WebNode& node) { | |
178 if (node.isTextNode()) | |
179 return node.nodeValue(); | |
180 | |
181 WebNode child = node.firstChild(); | |
182 | |
183 const int kChildSearchDepth = 10; | |
184 string16 node_text = FindChildTextInner(child, kChildSearchDepth); | |
185 TrimWhitespace(node_text, TRIM_ALL, &node_text); | |
186 return node_text; | |
187 } | |
188 | |
189 // Helper for |InferLabelForElement()| that infers a label, if possible, from | |
190 // a previous sibling of |element|, | |
191 // e.g. Some Text <input ...> | |
192 // or Some <span>Text</span> <input ...> | |
193 // or <p>Some Text</p><input ...> | |
194 // or <label>Some Text</label> <input ...> | |
195 // or Some Text <img><input ...> | |
196 // or <b>Some Text</b><br/> <input ...>. | |
197 string16 InferLabelFromPrevious(const WebFormControlElement& element) { | |
198 string16 inferred_label; | |
199 WebNode previous = element; | |
200 while (true) { | |
201 previous = previous.previousSibling(); | |
202 if (previous.isNull()) | |
203 break; | |
204 | |
205 // Skip over comments. | |
206 WebNode::NodeType node_type = previous.nodeType(); | |
207 if (node_type == WebNode::CommentNode) | |
208 continue; | |
209 | |
210 // Otherwise, only consider normal HTML elements and their contents. | |
211 if (node_type != WebNode::TextNode && | |
212 node_type != WebNode::ElementNode) | |
213 break; | |
214 | |
215 // A label might be split across multiple "lightweight" nodes. | |
216 // Coalesce any text contained in multiple consecutive | |
217 // (a) plain text nodes or | |
218 // (b) inline HTML elements that are essentially equivalent to text nodes. | |
219 CR_DEFINE_STATIC_LOCAL(WebString, kBold, ("b")); | |
220 CR_DEFINE_STATIC_LOCAL(WebString, kStrong, ("strong")); | |
221 CR_DEFINE_STATIC_LOCAL(WebString, kSpan, ("span")); | |
222 CR_DEFINE_STATIC_LOCAL(WebString, kFont, ("font")); | |
223 if (previous.isTextNode() || | |
224 HasTagName(previous, kBold) || HasTagName(previous, kStrong) || | |
225 HasTagName(previous, kSpan) || HasTagName(previous, kFont)) { | |
226 string16 value = FindChildText(previous); | |
227 // A text node's value will be empty if it is for a line break. | |
228 bool add_space = previous.isTextNode() && value.empty(); | |
229 inferred_label = | |
230 CombineAndCollapseWhitespace(value, inferred_label, add_space); | |
231 continue; | |
232 } | |
233 | |
234 // If we have identified a partial label and have reached a non-lightweight | |
235 // element, consider the label to be complete. | |
236 string16 trimmed_label; | |
237 TrimWhitespace(inferred_label, TRIM_ALL, &trimmed_label); | |
238 if (!trimmed_label.empty()) | |
239 break; | |
240 | |
241 // <img> and <br> tags often appear between the input element and its | |
242 // label text, so skip over them. | |
243 CR_DEFINE_STATIC_LOCAL(WebString, kImage, ("img")); | |
244 CR_DEFINE_STATIC_LOCAL(WebString, kBreak, ("br")); | |
245 if (HasTagName(previous, kImage) || HasTagName(previous, kBreak)) | |
246 continue; | |
247 | |
248 // We only expect <p> and <label> tags to contain the full label text. | |
249 CR_DEFINE_STATIC_LOCAL(WebString, kPage, ("p")); | |
250 CR_DEFINE_STATIC_LOCAL(WebString, kLabel, ("label")); | |
251 if (HasTagName(previous, kPage) || HasTagName(previous, kLabel)) | |
252 inferred_label = FindChildText(previous); | |
253 | |
254 break; | |
255 } | |
256 | |
257 TrimWhitespace(inferred_label, TRIM_ALL, &inferred_label); | |
258 return inferred_label; | |
259 } | |
260 | |
261 // Helper for |InferLabelForElement()| that infers a label, if possible, from | |
262 // enclosing list item, | |
263 // e.g. <li>Some Text<input ...><input ...><input ...></tr> | |
264 string16 InferLabelFromListItem(const WebFormControlElement& element) { | |
265 WebNode parent = element.parentNode(); | |
266 CR_DEFINE_STATIC_LOCAL(WebString, kListItem, ("li")); | |
267 while (!parent.isNull() && parent.isElementNode() && | |
268 !parent.to<WebElement>().hasTagName(kListItem)) { | |
269 parent = parent.parentNode(); | |
270 } | |
271 | |
272 if (!parent.isNull() && HasTagName(parent, kListItem)) | |
273 return FindChildText(parent); | |
274 | |
275 return string16(); | |
276 } | |
277 | |
278 // Helper for |InferLabelForElement()| that infers a label, if possible, from | |
279 // surrounding table structure, | |
280 // e.g. <tr><td>Some Text</td><td><input ...></td></tr> | |
281 // or <tr><th>Some Text</th><td><input ...></td></tr> | |
282 // or <tr><td><b>Some Text</b></td><td><b><input ...></b></td></tr> | |
283 // or <tr><th><b>Some Text</b></th><td><b><input ...></b></td></tr> | |
284 string16 InferLabelFromTableColumn(const WebFormControlElement& element) { | |
285 CR_DEFINE_STATIC_LOCAL(WebString, kTableCell, ("td")); | |
286 WebNode parent = element.parentNode(); | |
287 while (!parent.isNull() && parent.isElementNode() && | |
288 !parent.to<WebElement>().hasTagName(kTableCell)) { | |
289 parent = parent.parentNode(); | |
290 } | |
291 | |
292 if (parent.isNull()) | |
293 return string16(); | |
294 | |
295 // Check all previous siblings, skipping non-element nodes, until we find a | |
296 // non-empty text block. | |
297 string16 inferred_label; | |
298 WebNode previous = parent.previousSibling(); | |
299 CR_DEFINE_STATIC_LOCAL(WebString, kTableHeader, ("th")); | |
300 while (inferred_label.empty() && !previous.isNull()) { | |
301 if (HasTagName(previous, kTableCell) || HasTagName(previous, kTableHeader)) | |
302 inferred_label = FindChildText(previous); | |
303 | |
304 previous = previous.previousSibling(); | |
305 } | |
306 | |
307 return inferred_label; | |
308 } | |
309 | |
310 // Helper for |InferLabelForElement()| that infers a label, if possible, from | |
311 // surrounding table structure, | |
312 // e.g. <tr><td>Some Text</td></tr><tr><td><input ...></td></tr> | |
313 string16 InferLabelFromTableRow(const WebFormControlElement& element) { | |
314 CR_DEFINE_STATIC_LOCAL(WebString, kTableRow, ("tr")); | |
315 WebNode parent = element.parentNode(); | |
316 while (!parent.isNull() && parent.isElementNode() && | |
317 !parent.to<WebElement>().hasTagName(kTableRow)) { | |
318 parent = parent.parentNode(); | |
319 } | |
320 | |
321 if (parent.isNull()) | |
322 return string16(); | |
323 | |
324 // Check all previous siblings, skipping non-element nodes, until we find a | |
325 // non-empty text block. | |
326 string16 inferred_label; | |
327 WebNode previous = parent.previousSibling(); | |
328 while (inferred_label.empty() && !previous.isNull()) { | |
329 if (HasTagName(previous, kTableRow)) | |
330 inferred_label = FindChildText(previous); | |
331 | |
332 previous = previous.previousSibling(); | |
333 } | |
334 | |
335 return inferred_label; | |
336 } | |
337 | |
338 // Helper for |InferLabelForElement()| that infers a label, if possible, from | |
339 // a surrounding div table, | |
340 // e.g. <div>Some Text<span><input ...></span></div> | |
341 // e.g. <div>Some Text</div><div><input ...></div> | |
342 string16 InferLabelFromDivTable(const WebFormControlElement& element) { | |
343 WebNode node = element.parentNode(); | |
344 bool looking_for_parent = true; | |
345 | |
346 // Search the sibling and parent <div>s until we find a candidate label. | |
347 string16 inferred_label; | |
348 CR_DEFINE_STATIC_LOCAL(WebString, kDiv, ("div")); | |
349 CR_DEFINE_STATIC_LOCAL(WebString, kTable, ("table")); | |
350 CR_DEFINE_STATIC_LOCAL(WebString, kFieldSet, ("fieldset")); | |
351 while (inferred_label.empty() && !node.isNull()) { | |
352 if (HasTagName(node, kDiv)) { | |
353 looking_for_parent = false; | |
354 inferred_label = FindChildText(node); | |
355 } else if (looking_for_parent && | |
356 (HasTagName(node, kTable) || HasTagName(node, kFieldSet))) { | |
357 // If the element is in a table or fieldset, its label most likely is too. | |
358 break; | |
359 } | |
360 | |
361 if (node.previousSibling().isNull()) { | |
362 // If there are no more siblings, continue walking up the tree. | |
363 looking_for_parent = true; | |
364 } | |
365 | |
366 if (looking_for_parent) | |
367 node = node.parentNode(); | |
368 else | |
369 node = node.previousSibling(); | |
370 } | |
371 | |
372 return inferred_label; | |
373 } | |
374 | |
375 // Helper for |InferLabelForElement()| that infers a label, if possible, from | |
376 // a surrounding definition list, | |
377 // e.g. <dl><dt>Some Text</dt><dd><input ...></dd></dl> | |
378 // e.g. <dl><dt><b>Some Text</b></dt><dd><b><input ...></b></dd></dl> | |
379 string16 InferLabelFromDefinitionList(const WebFormControlElement& element) { | |
380 CR_DEFINE_STATIC_LOCAL(WebString, kDefinitionData, ("dd")); | |
381 WebNode parent = element.parentNode(); | |
382 while (!parent.isNull() && parent.isElementNode() && | |
383 !parent.to<WebElement>().hasTagName(kDefinitionData)) | |
384 parent = parent.parentNode(); | |
385 | |
386 if (parent.isNull() || !HasTagName(parent, kDefinitionData)) | |
387 return string16(); | |
388 | |
389 // Skip by any intervening text nodes. | |
390 WebNode previous = parent.previousSibling(); | |
391 while (!previous.isNull() && previous.isTextNode()) | |
392 previous = previous.previousSibling(); | |
393 | |
394 CR_DEFINE_STATIC_LOCAL(WebString, kDefinitionTag, ("dt")); | |
395 if (previous.isNull() || !HasTagName(previous, kDefinitionTag)) | |
396 return string16(); | |
397 | |
398 return FindChildText(previous); | |
399 } | |
400 | |
401 // Infers corresponding label for |element| from surrounding context in the DOM, | |
402 // e.g. the contents of the preceding <p> tag or text element. | |
403 string16 InferLabelForElement(const WebFormControlElement& element) { | |
404 string16 inferred_label = InferLabelFromPrevious(element); | |
405 if (!inferred_label.empty()) | |
406 return inferred_label; | |
407 | |
408 // If we didn't find a label, check for list item case. | |
409 inferred_label = InferLabelFromListItem(element); | |
410 if (!inferred_label.empty()) | |
411 return inferred_label; | |
412 | |
413 // If we didn't find a label, check for table cell case. | |
414 inferred_label = InferLabelFromTableColumn(element); | |
415 if (!inferred_label.empty()) | |
416 return inferred_label; | |
417 | |
418 // If we didn't find a label, check for table row case. | |
419 inferred_label = InferLabelFromTableRow(element); | |
420 if (!inferred_label.empty()) | |
421 return inferred_label; | |
422 | |
423 // If we didn't find a label, check for definition list case. | |
424 inferred_label = InferLabelFromDefinitionList(element); | |
425 if (!inferred_label.empty()) | |
426 return inferred_label; | |
427 | |
428 // If we didn't find a label, check for div table case. | |
429 return InferLabelFromDivTable(element); | |
430 } | |
431 | |
432 // Fills |option_strings| with the values of the <option> elements present in | |
433 // |select_element|. | |
434 void GetOptionStringsFromElement(const WebSelectElement& select_element, | |
435 std::vector<string16>* option_values, | |
436 std::vector<string16>* option_contents) { | |
437 DCHECK(!select_element.isNull()); | |
438 | |
439 option_values->clear(); | |
440 option_contents->clear(); | |
441 WebVector<WebElement> list_items = select_element.listItems(); | |
442 option_values->reserve(list_items.size()); | |
443 option_contents->reserve(list_items.size()); | |
444 for (size_t i = 0; i < list_items.size(); ++i) { | |
445 if (IsOptionElement(list_items[i])) { | |
446 const WebOptionElement option = list_items[i].toConst<WebOptionElement>(); | |
447 option_values->push_back(option.value()); | |
448 option_contents->push_back(option.text()); | |
449 } | |
450 } | |
451 } | |
452 | |
453 // The callback type used by |ForEachMatchingFormField()|. | |
454 typedef void (*Callback)(const FormFieldData&, | |
455 bool, /* is_initiating_element */ | |
456 WebKit::WebFormControlElement*); | |
457 | |
458 // For each autofillable field in |data| that matches a field in the |form|, | |
459 // the |callback| is invoked with the corresponding |form| field data. | |
460 void ForEachMatchingFormField(const WebFormElement& form_element, | |
461 const WebElement& initiating_element, | |
462 const FormData& data, | |
463 bool only_focusable_elements, | |
464 bool force_override, | |
465 Callback callback) { | |
466 std::vector<WebFormControlElement> control_elements; | |
467 ExtractAutofillableElements(form_element, autofill::REQUIRE_AUTOCOMPLETE, | |
468 &control_elements); | |
469 | |
470 if (control_elements.size() != data.fields.size()) { | |
471 // This case should be reachable only for pathological websites and tests, | |
472 // which add or remove form fields while the user is interacting with the | |
473 // Autofill popup. | |
474 return; | |
475 } | |
476 | |
477 // It's possible that the site has injected fields into the form after the | |
478 // page has loaded, so we can't assert that the size of the cached control | |
479 // elements is equal to the size of the fields in |form|. Fortunately, the | |
480 // one case in the wild where this happens, paypal.com signup form, the fields | |
481 // are appended to the end of the form and are not visible. | |
482 for (size_t i = 0; i < control_elements.size(); ++i) { | |
483 WebFormControlElement* element = &control_elements[i]; | |
484 | |
485 if (string16(element->nameForAutofill()) != data.fields[i].name) { | |
486 // This case should be reachable only for pathological websites, which | |
487 // rename form fields while the user is interacting with the Autofill | |
488 // popup. I (isherman) am not aware of any such websites, and so am | |
489 // optimistically including a NOTREACHED(). If you ever trip this check, | |
490 // please file a bug against me. | |
491 NOTREACHED(); | |
492 continue; | |
493 } | |
494 | |
495 bool is_initiating_element = (*element == initiating_element); | |
496 | |
497 // Only autofill empty fields and the field that initiated the filling, | |
498 // i.e. the field the user is currently editing and interacting with. | |
499 const WebInputElement* input_element = toWebInputElement(element); | |
500 if (!force_override && IsTextInput(input_element) && | |
501 !is_initiating_element && !input_element->value().isEmpty()) | |
502 continue; | |
503 | |
504 if (!element->isEnabled() || element->isReadOnly() || | |
505 (only_focusable_elements && !element->isFocusable())) | |
506 continue; | |
507 | |
508 callback(data.fields[i], is_initiating_element, element); | |
509 } | |
510 } | |
511 | |
512 // Sets the |field|'s value to the value in |data|. | |
513 // Also sets the "autofilled" attribute, causing the background to be yellow. | |
514 void FillFormField(const FormFieldData& data, | |
515 bool is_initiating_node, | |
516 WebKit::WebFormControlElement* field) { | |
517 // Nothing to fill. | |
518 if (data.value.empty()) | |
519 return; | |
520 | |
521 WebInputElement* input_element = toWebInputElement(field); | |
522 if (IsTextInput(input_element)) { | |
523 // If the maxlength attribute contains a negative value, maxLength() | |
524 // returns the default maxlength value. | |
525 input_element->setValue( | |
526 data.value.substr(0, input_element->maxLength()), true); | |
527 input_element->setAutofilled(true); | |
528 if (is_initiating_node) { | |
529 int length = input_element->value().length(); | |
530 input_element->setSelectionRange(length, length); | |
531 // Clear the current IME composition (the underline), if there is one. | |
532 input_element->document().frame()->unmarkText(); | |
533 } | |
534 } else if (IsSelectElement(*field)) { | |
535 WebSelectElement select_element = field->to<WebSelectElement>(); | |
536 if (select_element.value() != data.value) { | |
537 select_element.setValue(data.value); | |
538 select_element.dispatchFormControlChangeEvent(); | |
539 } | |
540 } else { | |
541 DCHECK(IsCheckableElement(input_element)); | |
542 input_element->setChecked(data.is_checked, true); | |
543 } | |
544 } | |
545 | |
546 // Sets the |field|'s "suggested" (non JS visible) value to the value in |data|. | |
547 // Also sets the "autofilled" attribute, causing the background to be yellow. | |
548 void PreviewFormField(const FormFieldData& data, | |
549 bool is_initiating_node, | |
550 WebKit::WebFormControlElement* field) { | |
551 // Nothing to preview. | |
552 if (data.value.empty()) | |
553 return; | |
554 | |
555 // Only preview input fields. Excludes checkboxes and radio buttons, as there | |
556 // is no provision for setSuggestedCheckedValue in WebInputElement. | |
557 WebInputElement* input_element = toWebInputElement(field); | |
558 if (!IsTextInput(input_element)) | |
559 return; | |
560 | |
561 // If the maxlength attribute contains a negative value, maxLength() | |
562 // returns the default maxlength value. | |
563 input_element->setSuggestedValue( | |
564 data.value.substr(0, input_element->maxLength())); | |
565 input_element->setAutofilled(true); | |
566 if (is_initiating_node) { | |
567 // Select the part of the text that the user didn't type. | |
568 input_element->setSelectionRange(input_element->value().length(), | |
569 input_element->suggestedValue().length()); | |
570 } | |
571 } | |
572 | |
573 std::string RetrievalMethodToString( | |
574 const autofill::WebElementDescriptor::RetrievalMethod& method) { | |
575 switch (method) { | |
576 case autofill::WebElementDescriptor::CSS_SELECTOR: | |
577 return "CSS_SELECTOR"; | |
578 case autofill::WebElementDescriptor::ID: | |
579 return "ID"; | |
580 case autofill::WebElementDescriptor::NONE: | |
581 return "NONE"; | |
582 } | |
583 NOTREACHED(); | |
584 return "UNKNOWN"; | |
585 } | |
586 | |
587 } // namespace | |
588 | |
589 namespace autofill { | |
590 | |
591 const size_t kMaxParseableFields = 100; | |
592 | |
593 // In HTML5, all text fields except password are text input fields to | |
594 // autocomplete. | |
595 bool IsTextInput(const WebInputElement* element) { | |
596 if (!element) | |
597 return false; | |
598 | |
599 return element->isTextField() && !element->isPasswordField(); | |
600 } | |
601 | |
602 bool IsSelectElement(const WebFormControlElement& element) { | |
603 // Is static for improving performance. | |
604 CR_DEFINE_STATIC_LOCAL(WebString, kSelectOne, ("select-one")); | |
605 return element.formControlType() == kSelectOne; | |
606 } | |
607 | |
608 bool IsCheckableElement(const WebInputElement* element) { | |
609 if (!element) | |
610 return false; | |
611 | |
612 return element->isCheckbox() || element->isRadioButton(); | |
613 } | |
614 | |
615 bool IsAutofillableInputElement(const WebInputElement* element) { | |
616 return IsTextInput(element) || IsCheckableElement(element); | |
617 } | |
618 | |
619 const string16 GetFormIdentifier(const WebFormElement& form) { | |
620 string16 identifier = form.name(); | |
621 CR_DEFINE_STATIC_LOCAL(WebString, kId, ("id")); | |
622 if (identifier.empty()) | |
623 identifier = form.getAttribute(kId); | |
624 | |
625 return identifier; | |
626 } | |
627 | |
628 bool ClickElement(const WebDocument& document, | |
629 const WebElementDescriptor& element_descriptor) { | |
630 WebString web_descriptor = WebString::fromUTF8(element_descriptor.descriptor); | |
631 WebKit::WebElement element; | |
632 | |
633 switch (element_descriptor.retrieval_method) { | |
634 case WebElementDescriptor::CSS_SELECTOR: { | |
635 WebExceptionCode ec = 0; | |
636 element = document.querySelector(web_descriptor, ec); | |
637 if (ec) | |
638 DVLOG(1) << "Query selector failed. Error code: " << ec << "."; | |
639 break; | |
640 } | |
641 case WebElementDescriptor::ID: | |
642 element = document.getElementById(web_descriptor); | |
643 break; | |
644 case WebElementDescriptor::NONE: | |
645 return true; | |
646 } | |
647 | |
648 if (element.isNull()) { | |
649 DVLOG(1) << "Could not find " | |
650 << element_descriptor.descriptor | |
651 << " by " | |
652 << RetrievalMethodToString(element_descriptor.retrieval_method) | |
653 << "."; | |
654 return false; | |
655 } | |
656 | |
657 element.simulateClick(); | |
658 return true; | |
659 } | |
660 | |
661 // Fills |autofillable_elements| with all the auto-fillable form control | |
662 // elements in |form_element|. | |
663 void ExtractAutofillableElements( | |
664 const WebFormElement& form_element, | |
665 RequirementsMask requirements, | |
666 std::vector<WebFormControlElement>* autofillable_elements) { | |
667 WebVector<WebFormControlElement> control_elements; | |
668 form_element.getFormControlElements(control_elements); | |
669 | |
670 autofillable_elements->clear(); | |
671 for (size_t i = 0; i < control_elements.size(); ++i) { | |
672 WebFormControlElement element = control_elements[i]; | |
673 if (!IsAutofillableElement(element)) | |
674 continue; | |
675 | |
676 if (requirements & REQUIRE_AUTOCOMPLETE) { | |
677 // TODO(jhawkins): WebKit currently doesn't handle the autocomplete | |
678 // attribute for select control elements, but it probably should. | |
679 WebInputElement* input_element = toWebInputElement(&control_elements[i]); | |
680 if (IsAutofillableInputElement(input_element) && | |
681 !SatisfiesRequireAutocomplete(*input_element)) | |
682 continue; | |
683 } | |
684 | |
685 autofillable_elements->push_back(element); | |
686 } | |
687 } | |
688 | |
689 void WebFormControlElementToFormField(const WebFormControlElement& element, | |
690 ExtractMask extract_mask, | |
691 FormFieldData* field) { | |
692 DCHECK(field); | |
693 DCHECK(!element.isNull()); | |
694 CR_DEFINE_STATIC_LOCAL(WebString, kAutocomplete, ("autocomplete")); | |
695 | |
696 // The label is not officially part of a WebFormControlElement; however, the | |
697 // labels for all form control elements are scraped from the DOM and set in | |
698 // WebFormElementToFormData. | |
699 field->name = element.nameForAutofill(); | |
700 field->form_control_type = UTF16ToUTF8(element.formControlType()); | |
701 field->autocomplete_attribute = | |
702 UTF16ToUTF8(element.getAttribute(kAutocomplete)); | |
703 if (field->autocomplete_attribute.size() > kMaxDataLength) { | |
704 // Discard overly long attribute values to avoid DOS-ing the browser | |
705 // process. However, send over a default string to indicate that the | |
706 // attribute was present. | |
707 field->autocomplete_attribute = "x-max-data-length-exceeded"; | |
708 } | |
709 | |
710 if (!IsAutofillableElement(element)) | |
711 return; | |
712 | |
713 const WebInputElement* input_element = toWebInputElement(&element); | |
714 if (IsAutofillableInputElement(input_element)) { | |
715 if (IsTextInput(input_element)) | |
716 field->max_length = input_element->maxLength(); | |
717 | |
718 field->is_autofilled = input_element->isAutofilled(); | |
719 field->is_focusable = input_element->isFocusable(); | |
720 field->should_autocomplete = input_element->autoComplete(); | |
721 field->is_checkable = IsCheckableElement(input_element); | |
722 } else if (extract_mask & EXTRACT_OPTIONS) { | |
723 // Set option strings on the field if available. | |
724 DCHECK(IsSelectElement(element)); | |
725 const WebSelectElement select_element = element.toConst<WebSelectElement>(); | |
726 GetOptionStringsFromElement(select_element, | |
727 &field->option_values, | |
728 &field->option_contents); | |
729 } | |
730 | |
731 if (!(extract_mask & EXTRACT_VALUE)) | |
732 return; | |
733 | |
734 string16 value; | |
735 if (IsAutofillableInputElement(input_element)) { | |
736 value = input_element->value(); | |
737 } else { | |
738 DCHECK(IsSelectElement(element)); | |
739 const WebSelectElement select_element = element.toConst<WebSelectElement>(); | |
740 value = select_element.value(); | |
741 | |
742 // Convert the |select_element| value to text if requested. | |
743 if (extract_mask & EXTRACT_OPTION_TEXT) { | |
744 WebVector<WebElement> list_items = select_element.listItems(); | |
745 for (size_t i = 0; i < list_items.size(); ++i) { | |
746 if (IsOptionElement(list_items[i])) { | |
747 const WebOptionElement option_element = | |
748 list_items[i].toConst<WebOptionElement>(); | |
749 if (option_element.value() == value) { | |
750 value = option_element.text(); | |
751 break; | |
752 } | |
753 } | |
754 } | |
755 } | |
756 } | |
757 | |
758 // Constrain the maximum data length to prevent a malicious site from DOS'ing | |
759 // the browser: http://crbug.com/49332 | |
760 if (value.size() > kMaxDataLength) | |
761 value = value.substr(0, kMaxDataLength); | |
762 | |
763 field->value = value; | |
764 } | |
765 | |
766 bool WebFormElementToFormData( | |
767 const WebKit::WebFormElement& form_element, | |
768 const WebKit::WebFormControlElement& form_control_element, | |
769 RequirementsMask requirements, | |
770 ExtractMask extract_mask, | |
771 FormData* form, | |
772 FormFieldData* field) { | |
773 CR_DEFINE_STATIC_LOCAL(WebString, kLabel, ("label")); | |
774 CR_DEFINE_STATIC_LOCAL(WebString, kFor, ("for")); | |
775 CR_DEFINE_STATIC_LOCAL(WebString, kHidden, ("hidden")); | |
776 | |
777 const WebFrame* frame = form_element.document().frame(); | |
778 if (!frame) | |
779 return false; | |
780 | |
781 if (requirements & REQUIRE_AUTOCOMPLETE && !form_element.autoComplete()) | |
782 return false; | |
783 | |
784 form->name = GetFormIdentifier(form_element); | |
785 form->method = form_element.method(); | |
786 form->origin = frame->document().url(); | |
787 form->action = frame->document().completeURL(form_element.action()); | |
788 form->user_submitted = form_element.wasUserSubmitted(); | |
789 | |
790 // If the completed URL is not valid, just use the action we get from | |
791 // WebKit. | |
792 if (!form->action.is_valid()) | |
793 form->action = GURL(form_element.action()); | |
794 | |
795 // A map from a FormFieldData's name to the FormFieldData itself. | |
796 std::map<string16, FormFieldData*> name_map; | |
797 | |
798 // The extracted FormFields. We use pointers so we can store them in | |
799 // |name_map|. | |
800 ScopedVector<FormFieldData> form_fields; | |
801 | |
802 WebVector<WebFormControlElement> control_elements; | |
803 form_element.getFormControlElements(control_elements); | |
804 | |
805 // A vector of bools that indicate whether each field in the form meets the | |
806 // requirements and thus will be in the resulting |form|. | |
807 std::vector<bool> fields_extracted(control_elements.size(), false); | |
808 | |
809 for (size_t i = 0; i < control_elements.size(); ++i) { | |
810 const WebFormControlElement& control_element = control_elements[i]; | |
811 | |
812 if (!IsAutofillableElement(control_element)) | |
813 continue; | |
814 | |
815 const WebInputElement* input_element = toWebInputElement(&control_element); | |
816 if (requirements & REQUIRE_AUTOCOMPLETE && | |
817 IsAutofillableInputElement(input_element) && | |
818 !SatisfiesRequireAutocomplete(*input_element)) | |
819 continue; | |
820 | |
821 // Create a new FormFieldData, fill it out and map it to the field's name. | |
822 FormFieldData* form_field = new FormFieldData; | |
823 WebFormControlElementToFormField(control_element, extract_mask, form_field); | |
824 form_fields.push_back(form_field); | |
825 // TODO(jhawkins): A label element is mapped to a form control element's id. | |
826 // field->name() will contain the id only if the name does not exist. Add | |
827 // an id() method to WebFormControlElement and use that here. | |
828 name_map[form_field->name] = form_field; | |
829 fields_extracted[i] = true; | |
830 } | |
831 | |
832 // If we failed to extract any fields, give up. Also, to avoid overly | |
833 // expensive computation, we impose a maximum number of allowable fields. | |
834 if (form_fields.empty() || form_fields.size() > kMaxParseableFields) | |
835 return false; | |
836 | |
837 // Loop through the label elements inside the form element. For each label | |
838 // element, get the corresponding form control element, use the form control | |
839 // element's name as a key into the <name, FormFieldData> map to find the | |
840 // previously created FormFieldData and set the FormFieldData's label to the | |
841 // label.firstChild().nodeValue() of the label element. | |
842 WebNodeList labels = form_element.getElementsByTagName(kLabel); | |
843 for (unsigned i = 0; i < labels.length(); ++i) { | |
844 WebLabelElement label = labels.item(i).to<WebLabelElement>(); | |
845 WebFormControlElement field_element = | |
846 label.correspondingControl().to<WebFormControlElement>(); | |
847 | |
848 string16 element_name; | |
849 if (field_element.isNull()) { | |
850 // Sometimes site authors will incorrectly specify the corresponding | |
851 // field element's name rather than its id, so we compensate here. | |
852 element_name = label.getAttribute(kFor); | |
853 } else if ( | |
854 !field_element.isFormControlElement() || | |
855 field_element.formControlType() == kHidden) { | |
856 continue; | |
857 } else { | |
858 element_name = field_element.nameForAutofill(); | |
859 } | |
860 | |
861 std::map<string16, FormFieldData*>::iterator iter = | |
862 name_map.find(element_name); | |
863 if (iter != name_map.end()) { | |
864 string16 label_text = FindChildText(label); | |
865 | |
866 // Concatenate labels because some sites might have multiple label | |
867 // candidates. | |
868 if (!iter->second->label.empty() && !label_text.empty()) | |
869 iter->second->label += ASCIIToUTF16(" "); | |
870 iter->second->label += label_text; | |
871 } | |
872 } | |
873 | |
874 // Loop through the form control elements, extracting the label text from | |
875 // the DOM. We use the |fields_extracted| vector to make sure we assign the | |
876 // extracted label to the correct field, as it's possible |form_fields| will | |
877 // not contain all of the elements in |control_elements|. | |
878 for (size_t i = 0, field_idx = 0; | |
879 i < control_elements.size() && field_idx < form_fields.size(); ++i) { | |
880 // This field didn't meet the requirements, so don't try to find a label | |
881 // for it. | |
882 if (!fields_extracted[i]) | |
883 continue; | |
884 | |
885 const WebFormControlElement& control_element = control_elements[i]; | |
886 if (form_fields[field_idx]->label.empty()) | |
887 form_fields[field_idx]->label = InferLabelForElement(control_element); | |
888 | |
889 if (field && form_control_element == control_element) | |
890 *field = *form_fields[field_idx]; | |
891 | |
892 ++field_idx; | |
893 } | |
894 | |
895 // Copy the created FormFields into the resulting FormData object. | |
896 for (ScopedVector<FormFieldData>::const_iterator iter = form_fields.begin(); | |
897 iter != form_fields.end(); ++iter) { | |
898 form->fields.push_back(**iter); | |
899 } | |
900 | |
901 return true; | |
902 } | |
903 | |
904 bool FindFormAndFieldForInputElement(const WebInputElement& element, | |
905 FormData* form, | |
906 FormFieldData* field, | |
907 RequirementsMask requirements) { | |
908 if (!IsAutofillableElement(element)) | |
909 return false; | |
910 | |
911 const WebFormElement form_element = element.form(); | |
912 if (form_element.isNull()) | |
913 return false; | |
914 | |
915 ExtractMask extract_mask = | |
916 static_cast<ExtractMask>(EXTRACT_VALUE | EXTRACT_OPTIONS); | |
917 return WebFormElementToFormData(form_element, | |
918 element, | |
919 requirements, | |
920 extract_mask, | |
921 form, | |
922 field); | |
923 } | |
924 | |
925 void FillForm(const FormData& form, const WebInputElement& element) { | |
926 WebFormElement form_element = element.form(); | |
927 if (form_element.isNull()) | |
928 return; | |
929 | |
930 ForEachMatchingFormField(form_element, | |
931 element, | |
932 form, | |
933 true, /* only_focusable_elements */ | |
934 false, /* dont force override */ | |
935 &FillFormField); | |
936 } | |
937 | |
938 void FillFormIncludingNonFocusableElements(const FormData& form_data, | |
939 const WebFormElement& form_element) { | |
940 if (form_element.isNull()) | |
941 return; | |
942 | |
943 ForEachMatchingFormField(form_element, | |
944 WebInputElement(), | |
945 form_data, | |
946 false, /* only_focusable_elements */ | |
947 true, /* force override */ | |
948 &FillFormField); | |
949 } | |
950 | |
951 void PreviewForm(const FormData& form, const WebInputElement& element) { | |
952 WebFormElement form_element = element.form(); | |
953 if (form_element.isNull()) | |
954 return; | |
955 | |
956 ForEachMatchingFormField(form_element, | |
957 element, | |
958 form, | |
959 true, /* only_focusable_elements */ | |
960 false, /* dont force override */ | |
961 &PreviewFormField); | |
962 } | |
963 | |
964 bool ClearPreviewedFormWithElement(const WebInputElement& element, | |
965 bool was_autofilled) { | |
966 WebFormElement form_element = element.form(); | |
967 if (form_element.isNull()) | |
968 return false; | |
969 | |
970 std::vector<WebFormControlElement> control_elements; | |
971 ExtractAutofillableElements(form_element, REQUIRE_AUTOCOMPLETE, | |
972 &control_elements); | |
973 for (size_t i = 0; i < control_elements.size(); ++i) { | |
974 // Only text input elements can be previewed. | |
975 WebInputElement* input_element = toWebInputElement(&control_elements[i]); | |
976 if (!IsTextInput(input_element)) | |
977 continue; | |
978 | |
979 // If the input element is not auto-filled, we did not preview it, so there | |
980 // is nothing to reset. | |
981 if (!input_element->isAutofilled()) | |
982 continue; | |
983 | |
984 // There might be unrelated elements in this form which have already been | |
985 // auto-filled. For example, the user might have already filled the address | |
986 // part of a form and now be dealing with the credit card section. We only | |
987 // want to reset the auto-filled status for fields that were previewed. | |
988 if (input_element->suggestedValue().isEmpty()) | |
989 continue; | |
990 | |
991 // Clear the suggested value. For the initiating node, also restore the | |
992 // original value. | |
993 input_element->setSuggestedValue(WebString()); | |
994 bool is_initiating_node = (element == *input_element); | |
995 if (is_initiating_node) | |
996 input_element->setAutofilled(was_autofilled); | |
997 else | |
998 input_element->setAutofilled(false); | |
999 | |
1000 // Clearing the suggested value in the focused node (above) can cause | |
1001 // selection to be lost. We force selection range to restore the text | |
1002 // cursor. | |
1003 if (is_initiating_node) { | |
1004 int length = input_element->value().length(); | |
1005 input_element->setSelectionRange(length, length); | |
1006 } | |
1007 } | |
1008 | |
1009 return true; | |
1010 } | |
1011 | |
1012 bool FormWithElementIsAutofilled(const WebInputElement& element) { | |
1013 WebFormElement form_element = element.form(); | |
1014 if (form_element.isNull()) | |
1015 return false; | |
1016 | |
1017 std::vector<WebFormControlElement> control_elements; | |
1018 ExtractAutofillableElements(form_element, REQUIRE_AUTOCOMPLETE, | |
1019 &control_elements); | |
1020 for (size_t i = 0; i < control_elements.size(); ++i) { | |
1021 WebInputElement* input_element = toWebInputElement(&control_elements[i]); | |
1022 if (!IsAutofillableInputElement(input_element)) | |
1023 continue; | |
1024 | |
1025 if (input_element->isAutofilled()) | |
1026 return true; | |
1027 } | |
1028 | |
1029 return false; | |
1030 } | |
1031 | |
1032 } // namespace autofill | |
OLD | NEW |