ideepu · ideepu · Apr 2, 2024 · Mar 28, 2024
diff --git a/codecov/coverage.py b/codecov/coverage.py
@@ -8,7 +8,7 @@
 import pathlib
 from collections.abc import Sequence
 
-from codecov import log, subprocess
+from codecov import log
 
 
 # The dataclasses in this module are accessible in the template, which is overridable by the user.
@@ -238,17 +238,6 @@ def get_diff_coverage_info(added_lines: dict[pathlib.Path, list[int]], coverage:
     )
 
 
-def get_added_lines(git: subprocess.Git, base_ref: str) -> dict[pathlib.Path, list[int]]:
-    # --unified=0 means we don't get any context lines for chunk, and we
-    # don't merge chunks. This means the headers that describe line number
-    # are always enough to derive what line numbers were added.
-    # TODO: check if it is possible to pull the diff changes from github API
-    # TODO: Since we don't want to checkout the code locally, it would be better to pull the diff from remote
-    git.fetch('origin', base_ref, '--depth=1000')
-    diff = git.diff('--unified=0', 'FETCH_HEAD', '--', '.')
-    return parse_diff_output(diff)
-
-
 def parse_diff_output(diff: str) -> dict[pathlib.Path, list[int]]:
     current_file: pathlib.Path | None = None
     added_filename_prefix = '+++ b/'
@@ -272,6 +261,8 @@ def parse_line_number_diff_line(line: str) -> Sequence[int]:
     Parse the "added" part of the line number diff text:
         @@ -60,0 +61 @@ def compute_files(  -> [61]
         @@ -60,0 +61,3 @@ def compute_files(  -> [61, 62, 63]
+
+    Github API returns default context lines 3 at start and end, so we can ignore them.
     """
     start, length = (int(i) for i in (line.split()[2][1:] + ',1').split(',')[:2])
-    return range(start, start + length)
+    return range(start + 3, start + length - 3)
diff --git a/codecov/github.py b/codecov/github.py
@@ -87,6 +87,21 @@ def get_pr_number(github: github_client.GitHub, config: settings.Config) -> int:
     )
 
 
+def get_pr_diff(github: github_client.GitHub, repository: str, pr_number: int) -> str:
+    try:
+        pull_request_diff = (
+            github.repos(repository)
+            .pulls(pr_number)
+            .get(use_text=True, headers={'Accept': 'application/vnd.github.v3.diff'})
+        )
+    except github_client.Forbidden as exc:
+        raise CannotGetPullRequest from exc
+    except github_client.NotFound as exc:
+        raise CannotGetPullRequest from exc
+
+    return pull_request_diff
+
+
 def post_comment(  # pylint: disable=too-many-arguments
     github: github_client.GitHub,
     me: str,

diff --git a/codecov/github_client.py b/codecov/github_client.py
@@ -48,9 +48,10 @@ def __init__(self, session: httpx.Client):
     def __getattr__(self, attr):
         return _Callable(self, f'/{attr}')
 
-    def _http(self, method, path, *, use_bytes=False, **kw):
+    def _http(self, method, path, *, use_bytes=False, use_text=False, **kw):
         _method = method.lower()
         requests_kwargs = {}
+        headers = kw.pop('headers', {})
         if _method == 'get' and kw:
             requests_kwargs = {'params': kw}
 
@@ -61,10 +62,13 @@ def _http(self, method, path, *, use_bytes=False, **kw):
             _method.upper(),
             path,
             timeout=TIMEOUT,
+            headers=headers,
             **requests_kwargs,
         )
         if use_bytes:
             contents = response.content
+        elif use_text:
+            contents = response.text
         else:
             contents = response_contents(response)
 

diff --git a/codecov/main.py b/codecov/main.py
@@ -5,17 +5,7 @@
 
 import httpx
 
-from codecov import (
-    coverage as coverage_module,
-    diff_grouper,
-    github,
-    github_client,
-    log,
-    log_utils,
-    settings,
-    subprocess,
-    template,
-)
+from codecov import coverage as coverage_module, diff_grouper, github, github_client, log, log_utils, settings, template
 
 
 def main():
@@ -33,9 +23,8 @@ def main():
             follow_redirects=True,
             headers={'Authorization': f'token {config.GITHUB_TOKEN}'},
         )
-        git = subprocess.Git()
 
-        exit_code = action(config=config, github_session=github_session, git=git)
+        exit_code = action(config=config, github_session=github_session)
         log.info('Ending action')
         sys.exit(exit_code)
 
@@ -46,7 +35,7 @@ def main():
         sys.exit(1)
 
 
-def action(config: settings.Config, github_session: httpx.Client, git: subprocess.Git) -> int:
+def action(config: settings.Config, github_session: httpx.Client) -> int:
     log.debug('Fetching Pull Request')
     gh = github_client.GitHub(session=github_session)
     try:
@@ -67,7 +56,6 @@ def action(config: settings.Config, github_session: httpx.Client, git: subproces
         config=config,
         gh=gh,
         repo_info=repo_info,
-        git=git,
         pr_number=pr_number,
     )
 
@@ -76,15 +64,13 @@ def process_pr(  # pylint: disable=too-many-locals
     config: settings.Config,
     gh: github_client.GitHub,
     repo_info: github.RepositoryInfo,
-    git: subprocess.Git,
     pr_number: int,
 ) -> int:
     log.info('Generating comment for PR')
-    _, coverage = coverage_module.get_coverage_info(
-        coverage_path=config.COVERAGE_PATH,
-    )
+    _, coverage = coverage_module.get_coverage_info(coverage_path=config.COVERAGE_PATH)
     base_ref = config.GITHUB_BASE_REF or repo_info.default_branch
-    added_lines = coverage_module.get_added_lines(git=git, base_ref=base_ref)
+    pr_diff = github.get_pr_diff(github=gh, repository=config.GITHUB_REPOSITORY, pr_number=pr_number)
+    added_lines = coverage_module.parse_diff_output(diff=pr_diff)
     diff_coverage = coverage_module.get_diff_coverage_info(coverage=coverage, added_lines=added_lines)
     marker = template.get_marker(marker_id=config.SUBPROJECT_ID)
 

diff --git a/codecov/subprocess.py b/codecov/subprocess.py