codedog-ai · Arcadia822 · Jul 21, 2023 · Jul 14, 2023 · Jul 21, 2023 · Jul 21, 2023
diff --git a/codedog/adapters/github_adapter.py b/codedog/adapters/github_adapter.py
@@ -38,8 +38,7 @@
 
 # used for github app
 github_app_id = env.get("GITHUB_APP_ID", 0)
-github_private_key = load_private_key(
-    env.get("GITHUB_PRIVATE_KEY_PATH", "/app/private_key.pem"))
+github_private_key = load_private_key(env.get("GITHUB_PRIVATE_KEY_PATH", "/app/private_key.pem"))
 
 
 issue_pattern = re.compile(r"#[0-9]+")
@@ -53,7 +52,7 @@ class GithubEvent(BaseModel):
     installation: dict = {}
 
 
-def handle_github_event(event: GithubEvent, local=False, **args) -> str:
+def handle_github_event(event: GithubEvent, local=False, **kwargs) -> str:
     # TODO: parse event related exception
     _event_filter(event)
 
@@ -64,40 +63,42 @@ def handle_github_event(event: GithubEvent, local=False, **args) -> str:
     assert repository_id
     assert pull_request_number
     # TODO: config
-    return handle_pull_request(repository_id, pull_request_number, installation_id, local, get_ttl_hash(120), **args)
+    return handle_pull_request(repository_id, pull_request_number, installation_id, local, get_ttl_hash(120), **kwargs)
 
 
 def get_github_client(installation_id: int):
     if installation_id is None or installation_id == 0:
         return default_gh
     jwt_token = get_jwt_token(github_private_key, github_app_id)
-    access_token = get_access_token_by_installation_id(
-        installation_id, jwt_token)
+    access_token = get_access_token_by_installation_id(installation_id, jwt_token)
     github_client = Github(access_token)
     return github_client
 
 
 @lru_cache()
-def handle_pull_request(repository_id: int, pull_request_number: int, installation_id: int,
-                        local=False, ttl_hash=None, **args):
+def handle_pull_request(
+    repository_id: int, pull_request_number: int, installation_id: int, local=False, ttl_hash=None, **kwargs
+):
     del ttl_hash
     logger.info(
         "Retrive pull request from Github",
-        extra={"github.repo.id": repository_id,
-               "github.pull.number": pull_request_number,
-               "github.installation_id": installation_id},
+        extra={
+            "github.repo.id": repository_id,
+            "github.pull.number": pull_request_number,
+            "github.installation_id": installation_id,
+        },
     )
-    github_client = get_github_client(installation_id)
+    github_client = default_gh if local else get_github_client(installation_id)
+
     pr = get_pr(github_client, repository_id, pull_request_number)
 
     changes = pr.changes
 
     callbacks = []
     if not local:
-        callbacks = [_comment_callback(github_client.get_repo(
-            repository_id).get_pull(pull_request_number))]
+        callbacks = [_comment_callback(github_client.get_repo(repository_id).get_pull(pull_request_number))]
 
-    thread = threading.Thread(target=asyncio.run, args=(_review_wrapper(pr, changes, callbacks, **args),))
+    thread = threading.Thread(target=asyncio.run, args=(_review_wrapper(pr, changes, callbacks, **kwargs),))
     thread.start()
 
     return "Review Submitted."
@@ -192,8 +193,8 @@ def get_potential_issue(repo: Repository, pull: GithubPullRequest) -> Issue:
     return issue
 
 
-async def _review_wrapper(pr: PullRequest, changes: list[Change], callbacks: list[callable] = [], **args):
-    review = Review(pr=pr, changes=changes, callbacks=callbacks, **args)
+async def _review_wrapper(pr: PullRequest, changes: list[Change], callbacks: list[callable] = [], **kwargs):
+    review = Review(pr=pr, changes=changes, callbacks=callbacks, **kwargs)
     await review.execute()
 
 

diff --git a/codedog/chains.py b/codedog/chains.py
@@ -6,20 +6,20 @@
 from langchain.output_parsers import OutputFixingParser, PydanticOutputParser
 
 from codedog.model import ChangeSummary
-from codedog.templates import grimoire_cn
-
-GRIMOIRE = grimoire_cn
+from codedog.templates import grimoire_cn, grimoire_en
 
 
 class Chains:
-    def __init__(self, llm: BaseChatModel):
-        self._review_fallback_prompt = PromptTemplate.from_template(GRIMOIRE.PR_CHANGE_REVIEW_FALLBACK_TEMPLATE)
-        self._summary_prompt = PromptTemplate.from_template(GRIMOIRE.PR_SUMMARIZE_TEMPLATE)
-        self._feedback_prompt = PromptTemplate.from_template(GRIMOIRE.PR_SIMPLE_FEEDBACK_TEMPLATE)
+    def __init__(self, llm: BaseChatModel, lang: str = "cn"):
+        grimoire = grimoire_cn if lang == "cn" else grimoire_en
+
+        self._review_fallback_prompt = PromptTemplate.from_template(grimoire.PR_CHANGE_REVIEW_FALLBACK_TEMPLATE)
+        self._summary_prompt = PromptTemplate.from_template(grimoire.PR_SUMMARIZE_TEMPLATE)
+        self._feedback_prompt = PromptTemplate.from_template(grimoire.PR_SIMPLE_FEEDBACK_TEMPLATE)
         self._raw_review_parser = PydanticOutputParser(pydantic_object=ChangeSummary)
         self._review_parser = OutputFixingParser.from_llm(llm=llm, parser=self._raw_review_parser)
         self._review_prompt = PromptTemplate(
-            template=GRIMOIRE.PR_CHANGE_REVIEW_TEMPLATE,
+            template=grimoire.PR_CHANGE_REVIEW_TEMPLATE,
             input_variables=["text", "name"],
             partial_variables={"format_instructions": self._raw_review_parser.get_format_instructions()},
         )
@@ -30,9 +30,9 @@ def __init__(self, llm: BaseChatModel):
         self._feedback_chain = LLMChain(llm=llm, prompt=self._feedback_prompt)
 
     @staticmethod
-    def init_chains():
+    def init_chains(lang: str = "cn"):
         llm = load_llm()
-        return Chains(llm=llm)
+        return Chains(llm=llm, lang=lang)
 
     @property
     def llm(self):

diff --git a/codedog/report.py b/codedog/report.py
diff --git a/codedog/review.py b/codedog/review.py
@@ -5,18 +5,16 @@
 import json
 import logging
 import time
+import traceback
 
 from langchain.callbacks import get_openai_callback
 from langchain.schema import OutputParserException
 
 from codedog.chains import Chains
 from codedog.model import Change, ChangeSummary, PullRequest
-from codedog.report import generate_change_summary, generate_feedback
-from codedog.templates import grimoire_cn, template_cn
+from codedog.templates import template_cn, template_en
 from codedog.version import VERSION
 
-GRIMOIRE = grimoire_cn
-TEMPLATE = template_cn
 logger = logging.getLogger(__name__)
 
 # TODO: unit test
@@ -30,11 +28,13 @@ def __init__(
         pr: PullRequest,
         changes: list[Change],
         callbacks: list | None = None,
-        chains: Chains = None,
+        lang: str = "cn",
         **kwargs,
     ):
-        self._chains: Chains = chains if isinstance(chains, Chains) else Chains.init_chains()
-
+        assert lang in ("cn", "en")
+        self.language = lang
+        self._chains = Chains.init_chains(lang=lang)
+        self._template = template_cn if lang == "cn" else template_en
         # --- data --------------------
         self._pr = pr
         self._changes: list[Change] = changes
@@ -84,7 +84,7 @@ async def execute(self) -> None:
 
             logger.info("Success code review %s", self.json_str())
         except Exception as ex:
-            logger.warn("Fail code review %s %s", ex, self.json_str())
+            logger.warn("Fail code review %s %s %s", ex, self.json_str(), traceback.format_exc().replace("\n", "\\n"))
 
     def print_report(self) -> None:
         print(self.report())
@@ -148,27 +148,100 @@ async def _feedback(self):
             self._meter_api_call_tokens(cb.total_tokens, cb.total_cost)
 
     def _generate_report(self) -> str:
-        header: str = TEMPLATE.REPORT_HEADER.format(
+        header: str = self._template.REPORT_HEADER.format(
             repo_name=self._pr.repository_name, pr_number=self._pr.pr_id, url=self._pr.url, version=VERSION
         )
 
-        telemetry: str = TEMPLATE.REPORT_TELEMETRY.format(
+        telemetry: str = self._template.REPORT_TELEMETRY.format(
             start_time=datetime.datetime.fromtimestamp(self._telemetry["start_time"]).strftime("%Y-%m-%d %H:%M:%S"),
             time_usage=int(self._telemetry["time_usage"]),
             files=self._telemetry["files"],
             tokens=self._telemetry["tokens"],
             cost=self._telemetry.get("cost", 0),
         )
 
-        summary: str = TEMPLATE.REPORT_PR_SUMMARY.format(
+        summary: str = self._template.REPORT_PR_SUMMARY.format(
             pr_summary=self._pr_summary,
-            pr_changes_summary=generate_change_summary(self._changes),
+            pr_changes_summary=self._generate_change_summary(),
         )
-        feedback: str = TEMPLATE.REPORT_FEEDBACK.format(feedback=generate_feedback(self._changes))
+        feedback: str = self._template.REPORT_FEEDBACK.format(feedback=self._generate_feedback())
 
         report = "\n".join([header, telemetry, summary, feedback])
         return report
 
+    def _generate_change_summary(self) -> str:
+        """format change summary
+
+        Args:
+            changes (list[Change]): pr changes and reviews
+        Returns:
+            str: format markdown table string of change summary
+        """
+        changes = self._changes
+        important_changes = []
+        housekeeping_changes = []
+
+        important_idx = 1
+        housekeeping_idx = 1
+        for change in changes:
+            file_name = change.file_name or ""
+            url = change.url or ""
+            summary = change.summary or ""
+
+            text = summary.replace("\n", "<br/>") if summary else ""
+            text_template: str = self._template.TABLE_LINE.format(file_name=file_name, url=url, text=text)
+
+            if not change.major:
+                text = text_template.format(idx=important_idx)
+                important_idx += 1
+                important_changes.append(text)
+            else:
+                text = text_template.format(idx=housekeeping_idx)
+                housekeeping_idx += 1
+                housekeeping_changes.append(text)
+
+        important_changes = "\n".join(important_changes) if important_changes else self._template.TABLE_LINE_NODATA
+        housekeeping_changes = (
+            "\n".join(housekeeping_changes) if housekeeping_changes else self._template.TABLE_LINE_NODATA
+        )
+        text = self._template.CHANGE_SUMMARY.format(
+            important_changes=important_changes, housekeeping_changes=housekeeping_changes
+        )
+        return text
+
+    def _generate_feedback(self) -> str:
+        """format feedback
+
+        Args:
+            changes (list[Change]): pr changes and reviews
+        Returns:
+            str: format markdown table string of feedback
+        """
+        changes = self._changes
+
+        texts = []
+
+        idx = 1
+        for change in changes:
+            file_name = change.file_name
+            url = change.url
+
+            feedback = change.feedback
+            if (
+                not feedback
+                or feedback in ("ok", "OK")
+                or (len(feedback) < 30 and "ok" in feedback.lower())  # 移除ok + 其他短语的回复
+            ):
+                continue
+
+            text = f"{self._template.T3_TITLE_LINE.format(idx=idx, file_name=file_name, url=url)}\n\n{feedback}"
+
+            texts.append(text)
+            idx += 1
+
+        concat_feedback_text = "\n\n".join(texts) if texts else self._template.TABLE_LINE_NODATA
+        return concat_feedback_text
+
     async def _execute_callback(self):
         if not self._callbacks:
             self.print_report()

diff --git a/codedog/server.py b/codedog/server.py
@@ -56,9 +56,7 @@ async def gitlab(event: GitlabEvent, url: str, token: str, source: str = "") ->
 
 
 @app.post("/v1/webhook/github", response_model=Response)
-async def github(
-    event: GithubEvent,
-) -> Response:
+async def github(event: GithubEvent, lang: str = "en") -> Response:
     """Github webhook.
 
     Args:
@@ -67,7 +65,7 @@ async def github(
         Response: message.
     """
     try:
-        message = handle_github_event(event)
+        message = handle_github_event(event, lang=lang)
     except CodedogError as e:
         return Response(message=e.message, code=e.code)
     except Exception: