Added coverage script and tests.

build/android/coverage.py

Index: build/android/coverage.py
diff --git a/build/android/coverage.py b/build/android/coverage.py
new file mode 100755
index 0000000000000000000000000000000000000000..df28198e13ee8755885834c2da240e4bf8eb3321
--- /dev/null
+++ b/build/android/coverage.py
@@ -0,0 +1,237 @@
+#!/usr/bin/python
+# Copyright 2015 The Chromium Authors. All rights reserved.
+# Use of this source code is governed by a BSD-style license that can be
+# found in the LICENSE file.
+
+"""Generates incremental code coverage reports for Java code in Chromium."""
+
+import os
+from lxml import html
+
+
+class LineCoverage(object):
+  """Coverage information about a single line of code."""
+
+  NOT_EXECUTABLE = -1
+  NOT_COVERED = 0
+  COVERED = 1
+  PARTIALLY_COVERED = 2
+
+  def __init__(self, lineno, source, covered_status, fractional_line_coverage):
+    """Initializes LineCoverage.
+
+    Args:
+      lineno: Integer line number.
+      source: A string containing the original line of source code.
+      covered_status: The covered status of the line.
+      fractional_line_coverage: The fractional value representing the fraction
+        of instructions executed for a given line of code. Should be a floating
+        point number between [0.0 - 1.0].

[mikecase (-- gone --), 2015/07/07 17:12:33]

s/"[0.0 - 1.0]"/"(0.0, 1.0)"

[estevenson1, 2015/07/07 23:54:55]

Done.

+    """
+    self.lineno = lineno
+    self.source = source
+    self.covered_status = covered_status
+    self.fractional_line_coverage = fractional_line_coverage
+
+
+class _EmmaHtmlParser(object):
+  """Encapsulates HTML file parsing operations.
+
+  This class contains all operations related to parsing HTML files that were
+  produced using the EMMA code coverage tool. It uses the lxml module for
+  parsing.
+
+  Example HTML:
+
+  To get package links:
+    <a href="_files/1.html">org.chromium.chrome</a>
+    This is returned by the selector |XPATH_SELECT_PACKAGE_ELEMENTS|.

[mikecase (-- gone --), 2015/07/07 17:12:32]

nit:

I would maybe just change ....

"To get package links:"
to 
"Package links:"
because of the way the comment is written.

[estevenson1, 2015/07/07 23:54:55]

Done.

+
+  To get class links:

[mikecase (-- gone --), 2015/07/07 17:12:32]

As above, maybe...
s/"To get class links:"/"Class links:"

[estevenson1, 2015/07/07 23:54:55]

Done, and fixed other occurrences.

+    <a href="1e.html">DoActivity.java</a>
+    This is returned by the selector |XPATH_SELECT_CLASS_ELEMENTS|.
+
+  To get coverage information:
+    <tr class="p">
+       <td class="l" title="78% line coverage (7 out of 9)">108</td>
+       <td title="78% line coverage (7 out of 9 instructions)">
+         if (index < 0 || index = mSelectors.size()) index = 0;</td>
+    </tr>
+    <tr>
+       <td class="l">109</td>
+       <td> </td>
+    </tr>
+    <tr class="c">
+       <td class="l">110</td>
+       <td>        if (mSelectors.get(index) != null) {</td>
+    </tr>
+    <tr class="z">
+       <td class="l">111</td>
+       <td>            for (int i = 0; i < mSelectors.size(); i++) {</td>
+    </tr>
+    Each <tr> element is returned by the selector |XPATH_SELECT_LOC|.
+
+    We can parse this to get:
+      1. Line number
+      2. Line of source code
+      3. Coverage status (c, z, or p)
+      4. Fractional coverage value (% out of 100 if PARTIALLY_COVERED)
+  """
+  # Selector to match all <a> elements within the rows that are in the table
+  # that displays all of the different packages.
+  _XPATH_SELECT_PACKAGE_ELEMENTS = '/html/body/table[4]/tr[*]/td/a'
+
+  # Selector to match all <a> elements within the rows that are in the table
+  # that displays all of the different packages within a class.
+  _XPATH_SELECT_CLASS_ELEMENTS = '/html/body/table[3]/tr[*]/td/a'
+
+  # Selector to match all <tr> elements within the table containing Java source
+  # code in an EMMA HTML file.
+  _XPATH_SELECT_LOC = '/html/body/table[4]/tr'
+
+  # Children of HTML elements are represented as a list in lxml. These constants
+  # represent list indices corresponding to relevant child elements.
+
+  # Technically both child 1 and 2 contain the percentage covered for a line.

[mikecase (-- gone --), 2015/07/07 17:12:33]

nit: clarify this comment. What does "technically both child 1 and 2" mean? If
it is that, literally both child 1 and 2 have the same information, I would just
choose 1 and delete this comment (only makes things more confusing)

[estevenson1, 2015/07/07 23:54:55]

Done.

+  _ELEMENT_PERCENT_COVERED = 1
+
+  # The second child contains the original line of source code.

[mikecase (-- gone --), 2015/07/07 17:12:32]

So this says "second child" (index 1) and above you call it "child 1" (also
index 1). I would stick to either calling them one way or the other.

[estevenson1, 2015/07/07 23:54:56]

Done.

+  _ELEMENT_CONTAINING_SOURCE_CODE = 1
+
+  # The first child contains the line number.
+  _ELEMENT_CONTAINING_LINENO = 0
+
+  # Maps CSS class names to corresponding coverage constants.
+  _CSS_TO_STATUS = {'c': LineCoverage.COVERED,

[mikecase (-- gone --), 2015/07/07 17:12:32]

nit: Would change formatting to be

_CSS_TO_STATUS = {
  'c': LineCoverage.COVERED,
}

[estevenson1, 2015/07/07 23:54:55]

Done.

+                    'p': LineCoverage.PARTIALLY_COVERED,
+                    'z': LineCoverage.NOT_COVERED}
+
+  # UTF-8 no break space.
+  _NO_BREAK_SPACE = '\xc2\xa0'
+
+  def __init__(self, emma_file_base_dir):
+    """Initializes _EmmaHtmlParser.
+
+    Args:
+      emma_file_base_dir: Path to the location where EMMA report files are
+        stored. Should be where index.html is stored.
+    """
+    self._base_dir = emma_file_base_dir
+    self._emma_files_path = os.path.join(self._base_dir, '_files')
+    self._index_path = os.path.join(self._base_dir, 'index.html')
+
+  def GetLineCoverage(self, emma_file_path):
+    """Returns a list of LineCoverage objects for the given EMMA HTML file.
+
+    Args:
+      emma_file_path: String representing the path to the EMMA HTML file.
+
+    Returns:
+      A list of LineCoverage objects.
+    """
+    def get_status(tr_element):
+      """Returns coverage status for a <tr> element containing coverage info."""
+      if 'class' not in tr_element.attrib:
+        status = LineCoverage.NOT_EXECUTABLE
+      else:
+        status = self._CSS_TO_STATUS.get(
+            tr_element.attrib['class'], LineCoverage.NOT_EXECUTABLE)
+      return status
+
+    def get_fractional_line_coverage(tr_element, status):
+      """Returns coverage value for a <tr> element containing coverage info."""
+      # If line is partially covered, parse the <td> tag to get the
+      # coverage percent.
+      if status == LineCoverage.PARTIALLY_COVERED:
+        title_attribute = (
+            tr_element[self._ELEMENT_PERCENT_COVERED].attrib['title'])
+        # Parse string that contains percent covered: "83% line coverage ,,,".
+        percent_covered = title_attribute.split('%')[0]
+        fractional_coverage_value = int(percent_covered) / 100.0
+      else:
+        fractional_coverage_value = 1.0
+      return fractional_coverage_value
+
+    def get_lineno(tr_element):
+      """Returns line number for a <tr> element containing coverage info."""
+      lineno_element = tr_element[self._ELEMENT_CONTAINING_LINENO]
+      # Handles oddly formatted HTML (where there is an extra <a> tag).
+      lineno = int(lineno_element.text or
+                   lineno_element[self._ELEMENT_CONTAINING_LINENO].text)
+      return lineno
+
+    def get_source_code(tr_element):
+      """Returns Java source for a <tr> element containing coverage info."""
+      raw_source = tr_element[self._ELEMENT_CONTAINING_SOURCE_CODE].text
+      utf8_source = raw_source.encode('UTF-8')
+      readable_source = utf8_source.replace(self._NO_BREAK_SPACE, ' ')
+      return readable_source
+
+    line_tr_elements = self._FindElements(emma_file_path,
+                                          _path=self._XPATH_SELECT_LOC)
+    line_coverage = []
+    for tr in line_tr_elements:
+      coverage_status = get_status(tr)
+      fractional_coverage = get_fractional_line_coverage(tr, coverage_status)
+      lineno = get_lineno(tr)
+      source = get_source_code(tr)
+      line = LineCoverage(lineno, source, coverage_status, fractional_coverage)
+      line_coverage.append(line)
+
+    return line_coverage
+
+  def GetPackageNameToEmmaFileDict(self):
+    """Returns a dict mapping Java packages to EMMA HTML coverage files.
+
+    Parses the EMMA index.html file to get a list of packages, then parses each
+    package HTML file to get a list of classes for that package, and creates
+    a dict with this info.
+
+    Returns:
+      A dict mapping string representation of Java packages (with class
+        names appended) to the corresponding file paths of EMMA HTML files.
+    """
+    # The <a> elements that contain each package name and the path of the file
+    # where all classes within said package are listed.
+    package_a_elements = self._FindElements(
+        self._index_path, _path=self._XPATH_SELECT_PACKAGE_ELEMENTS)
+    # Maps file path of package directory (EMMA generated) to package name.
+    # Ex. emma_dir/f.html: org.chromium.chrome.
+    package_links = {os.path.join(self._base_dir, link.attrib['href']):
+                     link.text
+                     for link in package_a_elements if 'href' in link.attrib}

[mikecase (-- gone --), 2015/07/07 17:12:33]

nit: Would prefer format....

package_links = {
    os.path.join(self._base_dir, link.attrib['href']): link.text
    for link in package_a_elements if 'href' in link.attrib
}

[estevenson1, 2015/07/07 23:54:55]

Done.

+
+    package_to_emma = {}
+    for package_emma_file_path, package_name in package_links.iteritems():
+      # The <a> elements that contain each class name in the current package and

[mikecase (-- gone --), 2015/07/07 17:12:33]

super nit: Kind of confusingly worked comments.

Maybe change...
" The <a> elements that contain..."
to
" These <a> elements contain..."

[estevenson1, 2015/07/07 23:54:55]

Done.

+      # the path of the file where the coverage info is stored for each class.

[mikecase (-- gone --), 2015/07/07 17:12:32]

Nit: capitalize "the"

+      coverage_file_a_elements = self._FindElements(
+          package_emma_file_path,
+          _path=self._XPATH_SELECT_CLASS_ELEMENTS)
+
+      for coverage_file_element in coverage_file_a_elements:
+        emma_file_path = os.path.join(self._emma_files_path,

[mikecase (-- gone --), 2015/07/07 17:12:32]

what is the difference between package_emma_file_path and emma_file_path?

[estevenson1, 2015/07/07 23:54:55]

package_emma_file_path is the path to the EMMA report html file that contains
all the links to different class coverage files within the specific package.
emma_file_path is simply a file path of an emma coverage file (i.e. the file
that stores coverage information for a Java source file). 

Renamed emma_file_path to emma_coverage_file_path.

+                                      coverage_file_element.attrib['href'])
+        full_package_name = package_name + '.' + coverage_file_element.text
+        package_to_emma[full_package_name] = emma_file_path
+
+    return package_to_emma
+
+  def _FindElements(self, file_path, **kwargs):

[mikecase (-- gone --), 2015/07/07 17:12:33]

Get rid of kwargs if you can. And just pass the xpath or something.

[estevenson1, 2015/07/07 23:54:55]

Done.

+    """Reads a HTML file and performs an XPath match.
+
+    Args:
+      file_path: String representing the path to the HTML file.
+      **kwargs: Keyword arguments for XPath match.
+
+    Returns:
+      A list of lxml.html.HtmlElements matching the given XPath selector.
+        Returns an empty list if there is no match.
+    """
+    try:
+      with open(file_path) as f:
+        file_contents = f.read().decode('ISO-8859-1')
+        root = html.fromstring(file_contents)
+        return root.xpath(**kwargs)
+    except IOError:
+      return []

[mikecase (-- gone --), 2015/07/07 17:12:33]

You should maybe just fail here? If this is expected to happen sometimes, you
should at least log an error.

[estevenson1, 2015/07/07 23:54:55]

Done.