Track number of translators (python-docs-translations#21)

* Track number of translators

* Handle no year in a translator record

* Get also from Git history

Author: m-aciek

completion.py

@@ -6,6 +6,9 @@
 import git
 from potodo import potodo
 
+import translators
+
+
 @cache
 def branches_from_devguide(devguide_dir: Path) -> list[str]:
     p = devguide_dir.joinpath('include/release-cycle.json')
@@ -15,20 +18,20 @@ def branches_from_devguide(devguide_dir: Path) -> list[str]:
     ]
 
 
-def get_completion(clones_dir: str, repo: str) -> float:
+def get_completion(clones_dir: str, repo: str) -> tuple[float, int]:
     clone_path = Path(clones_dir, repo)
     for branch in branches_from_devguide(Path(clones_dir, 'devguide')) + ['master']:
         try:
-            git.Repo.clone_from(
-                f'https://github.com/{repo}.git', clone_path, depth=1, branch=branch
-            )
+            git.Repo.clone_from(f'https://github.com/{repo}.git', clone_path, branch=branch)
         except git.GitCommandError:
             print(f'failed to clone {repo} {branch}')
+            translators_number = 0
             continue
         else:
+            translators_number = translators.get_number(clone_path)
             break
     with TemporaryDirectory() as tmpdir:
         completion = potodo.merge_and_scan_path(
             clone_path, pot_path=Path(clones_dir, 'cpython/Doc/build/gettext'), merge_path=Path(tmpdir), hide_reserved=False, api_url=''
         ).completion
-    return completion
+    return completion, translators_number

generate.py

@@ -32,11 +32,11 @@
     subprocess.run(['make', '-C', Path(clones_dir, 'cpython/Doc'), 'gettext'], check=True)
     for language, repo in repositories.get_languages_and_repos(devguide_dir):
         if repo:
-            completion_number = get_completion(clones_dir, repo)
+            completion_number, translators_number = get_completion(clones_dir, repo)
             visitors_number = visitors.get_number_of_visitors(language)
         else:
             completion_number, visitors_number = 0.0, 0
-        completion_progress.append((language, repo, completion_number, visitors_number))
+        completion_progress.append((language, repo, completion_number, translators_number, visitors_number))
         print(completion_progress[-1])
 
 template = Template(
@@ -53,11 +53,12 @@
 <tr>
   <th>language</th>
   <th><a href="https://plausible.io/data-policy#how-we-count-unique-users-without-cookies">visitors<a/></th>
+  <th>translators</th>
   <th>completion</th>
 </tr>
 </thead>
 <tbody>
-{% for language, repo, completion, visitors in completion_progress | sort(attribute=2) | reverse %}
+{% for language, repo, completion, translators, visitors in completion_progress | sort(attribute=2) | reverse %}
 <tr>
   {% if repo %}
   <td data-label="language">
@@ -70,9 +71,11 @@
       {{ '{:,}'.format(visitors) }}
     </a>
   </td>
+  <td data-label="translators">{{ '{:,}'.format(translators) }}</td>
   {% else %}
   <td data-label="language">{{ language }}</td>
   <td data-label="visitors">0</td>
+  <td data-label="translators">0</td>
   {% endif %}
   <td data-label="completion">
     <div class="progress-bar" style="width: {{ completion | round(2) }}%;">{{ completion | round(2) }}%</div>

translators.py

@@ -0,0 +1,30 @@
+from collections.abc import Generator
+from pathlib import Path
+
+from git import Repo
+from polib import pofile
+
+
+def get_number(path: Path) -> int:
+    from_headers = len(set(yield_from_headers(path)))
+    from_git_history = get_number_from_git_history(path)
+    return max(from_headers, from_git_history)
+
+def get_number_from_git_history(path: Path) -> int:
+    return len(Repo(path).git.shortlog('-s', 'HEAD').splitlines())
+
+def yield_from_headers(path: Path) -> Generator[str, None, None]:
+    for file in path.rglob('*.po'):
+        try:
+            header = pofile(file).header.splitlines()
+        except IOError:
+            continue
+        if 'Translators:' not in header:
+            continue
+        for translator_record in header[header.index('Translators:') + 1:]:
+            try:
+                translator, _year = translator_record.split(', ')
+            except ValueError:
+                yield translator_record
+            else:
+                yield translator