airbyte-internal-ops 0.8.0__py3-none-any.whl → 0.9.0__py3-none-any.whl

{airbyte_internal_ops-0.8.0.dist-info → airbyte_internal_ops-0.9.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: airbyte-internal-ops
-Version: 0.8.0
+Version: 0.9.0
 Summary: MCP and API interfaces that let the agents do the admin work
 Author-email: Aaron Steers <aj@airbyte.io>
 Keywords: admin,airbyte,api,mcp
@@ -32,6 +32,7 @@ Requires-Dist: jinja2<4.0,>=3.1.2
 Requires-Dist: markdown-it-py<3.0,>=2.2.0
 Requires-Dist: pydantic>=2.0.0
 Requires-Dist: pydash<8.0,>=6.0.2
+Requires-Dist: pygithub<3.0,>=2.0
 Requires-Dist: python-dotenv<2.0,>=1.0.0
 Requires-Dist: pyyaml<7.0,>=6.0.0
 Requires-Dist: requests<3.0,>=2.31.0

{airbyte_internal_ops-0.8.0.dist-info → airbyte_internal_ops-0.9.0.dist-info}/RECORD

@@ -4,7 +4,7 @@ airbyte_ops_mcp/constants.py,sha256=xU8ARMs7jG0-1hseKn3K1dmhyMOMM3JTetRZugLSKEo,
 airbyte_ops_mcp/docker_hub.py,sha256=qdOYpj2KOFOsEGsl2b2rcVPzyYDharOVM_lJxNTytds,5833
 airbyte_ops_mcp/gcp_auth.py,sha256=i0cm1_xX4fj_31iKlfARpNvTaSr85iGTSw9KMf4f4MU,7206
 airbyte_ops_mcp/github_actions.py,sha256=FSi_tjS9TbwRVp8dwlDZhFOi7lJXEZQLhPm2KpcjNlY,7022
-airbyte_ops_mcp/github_api.py,sha256=ezpMR1vjqQ-1f5yOLBVbxW70OPtUferl1uA0u_gUVo8,12733
+airbyte_ops_mcp/github_api.py,sha256=Hz3ZCGfLfnfSsytV4zpAFS6q3u0Xh8hUhSSRAwjf6G4,14398
 airbyte_ops_mcp/metadata_validator.py,sha256=kRCr3uzZmFLv4gCzY4KzLrApycqoE1YgDRpld04EBaU,15002
 airbyte_ops_mcp/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 airbyte_ops_mcp/telemetry.py,sha256=GUQX-oD6besFl-kAyycci1HCk1jov_JFwUAqKVN0yKE,5230
@@ -218,6 +218,7 @@ airbyte_ops_mcp/_legacy/airbyte_ci/metadata_service/templates/connector_registry
 airbyte_ops_mcp/_legacy/airbyte_ci/metadata_service/templates/render.py,sha256=gX5Fk5HJ8a_gnZ7N2rgFL8aKURHemWSKd_Z5RBISkpI,2831
 airbyte_ops_mcp/airbyte_repo/__init__.py,sha256=3yEpCOop1h33UuCtU7-1UB3kun2QsxtgvsfVEpuqN6s,1572
 airbyte_ops_mcp/airbyte_repo/bump_version.py,sha256=rtXYv105BhElj7twQmy20VjNsj6vCb0Ulrzmc4Z-1cE,15384
+airbyte_ops_mcp/airbyte_repo/changelog_fix.py,sha256=sT94FkH5YuPc_NZCoMr9IUydcD4CO-i5ifGm8FOuI9g,13890
 airbyte_ops_mcp/airbyte_repo/list_connectors.py,sha256=UI4RKMfFWOXHBtvh7TIw1v0BnV4rQQGuUgh8e9yvpos,19907
 airbyte_ops_mcp/airbyte_repo/utils.py,sha256=TXlOAfhiu_hVRNjCxB4PRPVDhTWCU5lYmgqz4QG_-EA,3201
 airbyte_ops_mcp/cli/__init__.py,sha256=XpL7FyVfgabfBF2JR7u7NwJ2krlYqjd_OwLcWf-Xc7s,114
@@ -226,8 +227,8 @@ airbyte_ops_mcp/cli/_shared.py,sha256=jg-xMyGzTCGPqKd8VTfE_3kGPIyO_3Kx5sQbG4rPc0
 airbyte_ops_mcp/cli/app.py,sha256=xxlyJzlFvZuy-qKAhTPPmzj-t0o7MSqQ7aPHonfT1vU,790
 airbyte_ops_mcp/cli/cloud.py,sha256=jac0FcG3UvNNeNNhqZR04uY9BCyVvins1kjO0LRZ86Y,45861
 airbyte_ops_mcp/cli/gh.py,sha256=koJPu0MDB6AW7mJq2z4dZV65ofvsZTkqoeitGF8KJR8,5364
-airbyte_ops_mcp/cli/local.py,sha256=57F_FrC8oEEa9eb85sptJOSDncS_ItlWtnO6DcJomKk,20874
-airbyte_ops_mcp/cli/registry.py,sha256=L4nDKhlegr31gSE-GUvDFSq10KgDz5kJuZXgLIxYIyg,9785
+airbyte_ops_mcp/cli/local.py,sha256=kJe3rO6AWdXyjvrlcyP5MC_wIGu9cm1DaCNuO2gxkhU,40806
+airbyte_ops_mcp/cli/registry.py,sha256=CVm15_6YWpLD99DZ8f8y68npXJw01Ckde7Ty3o89YNE,16631
 airbyte_ops_mcp/cloud_admin/__init__.py,sha256=cqE96Q10Kp6elhH9DAi6TVsIwSUy3sooDLLrxTaktGk,816
 airbyte_ops_mcp/cloud_admin/api_client.py,sha256=pzdnZpIxOaXJsF_vQuw9_goiDbqwgEQxFwZwsoupY40,57459
 airbyte_ops_mcp/cloud_admin/auth.py,sha256=qE2Aqe0qbZB755KscL65s54Jz78-F-X5a8fXKsrYEOQ,3749
@@ -283,8 +284,10 @@ airbyte_ops_mcp/prod_db_access/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMp
 airbyte_ops_mcp/prod_db_access/queries.py,sha256=PyZ0HU-E6jDhSgQ94XRz4wzh61vCr0ACxfMoGjeTKUU,22312
 airbyte_ops_mcp/prod_db_access/sql.py,sha256=E7HQratZC5LbCXdvpQzZRd09SmCVnY_4Ru-HluWaQ_c,38308
 airbyte_ops_mcp/registry/__init__.py,sha256=iEaPlt9GrnlaLbc__98TguNeZG8wuQu7S-_2QkhHcbA,858
+airbyte_ops_mcp/registry/_gcs_util.py,sha256=3dL7iqpyoSPfl2nGBPp73TqrqGWI4Ks6egKpQXDnw24,2838
+airbyte_ops_mcp/registry/connector_stubs.py,sha256=pBzaQIOgCfBvHuy3_N45amU9Do2ALDpU8q5HnZySPTQ,5807
 airbyte_ops_mcp/registry/models.py,sha256=B4L4TKr52wo0xs0CqvCBrpowqjShzVnZ5eTr2-EyhNs,2346
-airbyte_ops_mcp/registry/publish.py,sha256=VoPxsM2_0zJ829orzCRN-kjgcJtuBNyXgW4I9J680ro,12717
+airbyte_ops_mcp/registry/publish.py,sha256=BHOvj7qximKytGKTXMXCO6_1yTkmSvUije5jHKpQj64,12640
 airbyte_ops_mcp/regression_tests/__init__.py,sha256=qdveN866w50VmiDmTD0IW8Buo4UUn_1m9mKNS6hgj5U,1092
 airbyte_ops_mcp/regression_tests/cdk_secrets.py,sha256=iRjqqBS96KZoswfgT7ju-pE_pfbYoDy4PfrK-K8uyYs,3204
 airbyte_ops_mcp/regression_tests/ci_output.py,sha256=os69gcEhbomrGRAfnwPYAXhsAiJSWOuyW1gWZPx9-hw,15498
@@ -305,7 +308,7 @@ airbyte_ops_mcp/regression_tests/regression/comparators.py,sha256=MJkLZEKHivgrG0
 airbyte_ops_mcp/regression_tests/validation/__init__.py,sha256=MBEwGOoNuqT4_oCahtoK62OKWIjUCfWa7vZTxNj_0Ek,1532
 airbyte_ops_mcp/regression_tests/validation/catalog_validators.py,sha256=jqqVAMOk0mtdPgwu4d0hA0ZEjtsNh5gapvGydRv3_qk,12553
 airbyte_ops_mcp/regression_tests/validation/record_validators.py,sha256=RjauAhKWNwxMBTu0eNS2hMFNQVs5CLbQU51kp6FOVDk,7432
-airbyte_internal_ops-0.8.0.dist-info/METADATA,sha256=b9pAOYSCdbb5hYDF-Wuk8wWG1r2uh7-Su-8wzwPOE_s,5964
-airbyte_internal_ops-0.8.0.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
-airbyte_internal_ops-0.8.0.dist-info/entry_points.txt,sha256=WxP0l7bRFss4Cr5uQqVj9mTEKwnRKouNuphXQF0lotA,171
-airbyte_internal_ops-0.8.0.dist-info/RECORD,,
+airbyte_internal_ops-0.9.0.dist-info/METADATA,sha256=bE6ARIIiGF7okmQ779VU9T5KgHrX-2cdsv4wOtianI8,5998
+airbyte_internal_ops-0.9.0.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
+airbyte_internal_ops-0.9.0.dist-info/entry_points.txt,sha256=WxP0l7bRFss4Cr5uQqVj9mTEKwnRKouNuphXQF0lotA,171
+airbyte_internal_ops-0.9.0.dist-info/RECORD,,

airbyte_ops_mcp/airbyte_repo/changelog_fix.py

@@ -0,0 +1,437 @@
+# Copyright (c) 2025 Airbyte, Inc., all rights reserved.
+"""Changelog date fixing and checking utilities for Airbyte connectors.
+
+This module provides functionality to fix and check changelog entry dates by looking up
+the actual PR merge dates from GitHub.
+"""
+
+from __future__ import annotations
+
+import datetime
+import re
+from dataclasses import dataclass
+from enum import StrEnum
+from pathlib import Path
+
+from airbyte_ops_mcp.airbyte_repo.bump_version import (
+    AIRBYTE_GITHUB_REPO,
+    get_connector_doc_path,
+)
+from airbyte_ops_mcp.airbyte_repo.list_connectors import get_all_connectors
+from airbyte_ops_mcp.github_api import GitHubAPIError, get_pr_merge_date
+
+
+class ChangelogIssueType(StrEnum):
+    """Types of changelog issues that can be detected."""
+
+    PR_MISMATCH = "pr_mismatch"
+
+
+@dataclass
+class ChangelogDateFix:
+    """A single changelog date fix."""
+
+    line_number: int
+    version: str
+    pr_number: int
+    old_date: datetime.date
+    new_date: datetime.date
+
+    @property
+    def changed(self) -> bool:
+        """Return True if the date was changed."""
+        return self.old_date != self.new_date
+
+
+@dataclass
+class ChangelogIssue:
+    """A single changelog issue found during checking."""
+
+    line_number: int
+    version: str
+    issue_type: ChangelogIssueType
+    message: str
+
+
+@dataclass
+class ChangelogCheckResult:
+    """Result of checking changelog for a connector."""
+
+    connector: str
+    doc_path: Path | None
+    date_issues: list[ChangelogDateFix]
+    pr_mismatch_warnings: list[ChangelogIssue]
+    errors: list[str]
+
+    @property
+    def has_issues(self) -> bool:
+        """Return True if any issues were found."""
+        return (
+            any(fix.changed for fix in self.date_issues)
+            or len(self.pr_mismatch_warnings) > 0
+        )
+
+    @property
+    def issue_count(self) -> int:
+        """Return the total number of issues found."""
+        return sum(1 for fix in self.date_issues if fix.changed) + len(
+            self.pr_mismatch_warnings
+        )
+
+
+@dataclass
+class ChangelogFixResult:
+    """Result of fixing changelog dates for a connector."""
+
+    connector: str
+    doc_path: Path | None
+    fixes: list[ChangelogDateFix]
+    warnings: list[ChangelogIssue]
+    errors: list[str]
+    dry_run: bool
+
+    @property
+    def has_changes(self) -> bool:
+        """Return True if any dates were changed."""
+        return any(fix.changed for fix in self.fixes)
+
+    @property
+    def changed_count(self) -> int:
+        """Return the number of dates that were changed."""
+        return sum(1 for fix in self.fixes if fix.changed)
+
+
+def _parse_changelog_entries(
+    content: str,
+    github_repo: str = AIRBYTE_GITHUB_REPO,
+) -> list[tuple[int, str, str, int, int, str]]:
+    """Parse changelog entries from markdown content.
+
+    Args:
+        content: The markdown content of the documentation file.
+        github_repo: GitHub repository for PR links.
+
+    Returns:
+        List of tuples: (line_number, version, date_str, displayed_pr_number, url_pr_number, full_line)
+    """
+    # Regex to parse changelog table rows in the format:
+    # | version | date | [pr_num](url) | comment |
+    changelog_entry_re = (
+        # Match table row start and capture semantic version (e.g., "1.2.3")
+        r"^\| *(?P<version>[0-9]+\.[0-9]+\.[0-9]+) *\| *"
+        # Capture date in ISO format YYYY-MM-DD
+        r"(?P<date>[0-9]{4}-[0-9]{2}-[0-9]{2}) *\| *"
+        # Capture displayed PR number (may be in brackets as markdown link)
+        r"\[?(?P<displayed_pr>[0-9]+)\]?\(https://github.com/"
+        # GitHub repo portion (escaped to handle special chars)
+        + re.escape(github_repo)
+        # Capture PR number from URL path
+        + r"/pull/(?P<url_pr>[0-9]+)\) *\| *"
+        # Capture comment text until end of row
+        r"(?P<comment>.*?) *\| *$"
+    )
+
+    entries = []
+    lines = content.splitlines()
+
+    for line_num, line in enumerate(lines, start=1):
+        match = re.match(changelog_entry_re, line)
+        if match:
+            version = match.group("version")
+            date_str = match.group("date")
+            displayed_pr = int(match.group("displayed_pr"))
+            url_pr = int(match.group("url_pr"))
+            entries.append((line_num, version, date_str, displayed_pr, url_pr, line))
+
+    return entries
+
+
+def check_changelog(
+    repo_path: str | Path,
+    connector_name: str,
+    lookback_days: int | None = None,
+    github_repo: str = AIRBYTE_GITHUB_REPO,
+    token: str | None = None,
+) -> ChangelogCheckResult:
+    """Check changelog for issues (incorrect dates, mismatched PR numbers).
+
+    Args:
+        repo_path: Path to the Airbyte monorepo.
+        connector_name: Technical name of the connector (e.g., "source-github").
+        lookback_days: Only check entries with dates within this many days. None for all.
+        github_repo: GitHub repository for PR links.
+        token: GitHub API token. If None, will be resolved from environment.
+
+    Returns:
+        ChangelogCheckResult with details of any issues found.
+    """
+    repo_path = Path(repo_path)
+    date_issues: list[ChangelogDateFix] = []
+    pr_mismatch_warnings: list[ChangelogIssue] = []
+    errors: list[str] = []
+
+    doc_path = get_connector_doc_path(repo_path, connector_name)
+    if doc_path is None or not doc_path.exists():
+        return ChangelogCheckResult(
+            connector=connector_name,
+            doc_path=None,
+            date_issues=[],
+            pr_mismatch_warnings=[],
+            errors=[f"Documentation file not found for {connector_name}"],
+        )
+
+    content = doc_path.read_text()
+    entries = _parse_changelog_entries(content, github_repo)
+
+    if not entries:
+        return ChangelogCheckResult(
+            connector=connector_name,
+            doc_path=doc_path,
+            date_issues=[],
+            pr_mismatch_warnings=[],
+            errors=[],
+        )
+
+    owner, repo = github_repo.split("/")
+    cutoff_date = None
+    if lookback_days is not None:
+        cutoff_date = datetime.date.today() - datetime.timedelta(days=lookback_days)
+
+    for line_num, version, date_str, displayed_pr, url_pr, _full_line in entries:
+        entry_date = datetime.datetime.strptime(date_str, "%Y-%m-%d").date()
+
+        # Skip entries older than the lookback threshold
+        if cutoff_date is not None and entry_date < cutoff_date:
+            continue
+
+        if displayed_pr != url_pr:
+            pr_mismatch_warnings.append(
+                ChangelogIssue(
+                    line_number=line_num,
+                    version=version,
+                    issue_type=ChangelogIssueType.PR_MISMATCH,
+                    message=f"Displayed PR number ({displayed_pr}) does not match URL PR number ({url_pr})",
+                )
+            )
+
+        merge_date = None
+        try:
+            merge_date = get_pr_merge_date(owner, repo, url_pr, token)
+        except GitHubAPIError as e:
+            errors.append(f"Failed to fetch PR {url_pr}: {e}")
+            continue
+
+        if merge_date is None:
+            errors.append(f"PR {url_pr} is not merged")
+            continue
+
+        date_issues.append(
+            ChangelogDateFix(
+                line_number=line_num,
+                version=version,
+                pr_number=url_pr,
+                old_date=entry_date,
+                new_date=merge_date,
+            )
+        )
+
+    return ChangelogCheckResult(
+        connector=connector_name,
+        doc_path=doc_path,
+        date_issues=date_issues,
+        pr_mismatch_warnings=pr_mismatch_warnings,
+        errors=errors,
+    )
+
+
+def fix_changelog_dates(
+    repo_path: str | Path,
+    connector_name: str,
+    dry_run: bool = False,
+    lookback_days: int | None = None,
+    github_repo: str = AIRBYTE_GITHUB_REPO,
+    token: str | None = None,
+) -> ChangelogFixResult:
+    """Fix changelog dates for a connector by looking up PR merge dates.
+
+    Args:
+        repo_path: Path to the Airbyte monorepo.
+        connector_name: Technical name of the connector (e.g., "source-github").
+        dry_run: If True, don't actually modify the file.
+        lookback_days: Only fix entries with dates within this many days. None for all.
+        github_repo: GitHub repository for PR links.
+        token: GitHub API token. If None, will be resolved from environment.
+
+    Returns:
+        ChangelogFixResult with details of the fixes applied.
+    """
+    repo_path = Path(repo_path)
+    fixes: list[ChangelogDateFix] = []
+    warnings: list[ChangelogIssue] = []
+    errors: list[str] = []
+
+    doc_path = get_connector_doc_path(repo_path, connector_name)
+    if doc_path is None or not doc_path.exists():
+        return ChangelogFixResult(
+            connector=connector_name,
+            doc_path=None,
+            fixes=[],
+            warnings=[],
+            errors=[f"Documentation file not found for {connector_name}"],
+            dry_run=dry_run,
+        )
+
+    content = doc_path.read_text()
+    entries = _parse_changelog_entries(content, github_repo)
+
+    if not entries:
+        return ChangelogFixResult(
+            connector=connector_name,
+            doc_path=doc_path,
+            fixes=[],
+            warnings=[],
+            errors=[],
+            dry_run=dry_run,
+        )
+
+    owner, repo = github_repo.split("/")
+    lines = content.splitlines()
+    # Track which lines need to be modified: line_num (1-indexed) -> new_line
+    line_replacements: dict[int, str] = {}
+    cutoff_date = None
+    if lookback_days is not None:
+        cutoff_date = datetime.date.today() - datetime.timedelta(days=lookback_days)
+
+    for line_num, version, date_str, displayed_pr, url_pr, full_line in entries:
+        entry_date = datetime.datetime.strptime(date_str, "%Y-%m-%d").date()
+
+        # Skip entries older than the lookback threshold
+        if cutoff_date is not None and entry_date < cutoff_date:
+            continue
+
+        if displayed_pr != url_pr:
+            warnings.append(
+                ChangelogIssue(
+                    line_number=line_num,
+                    version=version,
+                    issue_type=ChangelogIssueType.PR_MISMATCH,
+                    message=f"Displayed PR number ({displayed_pr}) does not match URL PR number ({url_pr})",
+                )
+            )
+
+        merge_date = None
+        try:
+            merge_date = get_pr_merge_date(owner, repo, url_pr, token)
+        except GitHubAPIError as e:
+            errors.append(f"Failed to fetch PR {url_pr}: {e}")
+            continue
+
+        if merge_date is None:
+            errors.append(f"PR {url_pr} is not merged")
+            continue
+
+        fix = ChangelogDateFix(
+            line_number=line_num,
+            version=version,
+            pr_number=url_pr,
+            old_date=entry_date,
+            new_date=merge_date,
+        )
+        fixes.append(fix)
+
+        if fix.changed:
+            new_line = full_line.replace(date_str, merge_date.strftime("%Y-%m-%d"))
+            line_replacements[line_num] = new_line
+
+    # Apply line replacements by line number to avoid issues with duplicate lines
+    if line_replacements:
+        for line_num, new_line in line_replacements.items():
+            lines[line_num - 1] = new_line  # Convert 1-indexed to 0-indexed
+        new_content = "\n".join(lines)
+        if content.endswith("\n"):
+            new_content += "\n"
+    else:
+        new_content = content
+
+    if not dry_run and new_content != content:
+        doc_path.write_text(new_content)
+
+    return ChangelogFixResult(
+        connector=connector_name,
+        doc_path=doc_path,
+        fixes=fixes,
+        warnings=warnings,
+        errors=errors,
+        dry_run=dry_run,
+    )
+
+
+def check_all_changelogs(
+    repo_path: str | Path,
+    lookback_days: int | None = None,
+    github_repo: str = AIRBYTE_GITHUB_REPO,
+    token: str | None = None,
+) -> list[ChangelogCheckResult]:
+    """Check changelogs for all connectors in the repository.
+
+    Args:
+        repo_path: Path to the Airbyte monorepo.
+        lookback_days: Only check entries with dates within this many days. None for all.
+        github_repo: GitHub repository for PR links.
+        token: GitHub API token. If None, will be resolved from environment.
+
+    Returns:
+        List of ChangelogCheckResult for each connector processed.
+    """
+    repo_path = Path(repo_path)
+    connectors = get_all_connectors(repo_path)
+
+    results = []
+    for connector_name in sorted(connectors):
+        result = check_changelog(
+            repo_path=repo_path,
+            connector_name=connector_name,
+            lookback_days=lookback_days,
+            github_repo=github_repo,
+            token=token,
+        )
+        results.append(result)
+
+    return results
+
+
+def fix_all_changelog_dates(
+    repo_path: str | Path,
+    dry_run: bool = False,
+    lookback_days: int | None = None,
+    github_repo: str = AIRBYTE_GITHUB_REPO,
+    token: str | None = None,
+) -> list[ChangelogFixResult]:
+    """Fix changelog dates for all connectors in the repository.
+
+    Args:
+        repo_path: Path to the Airbyte monorepo.
+        dry_run: If True, don't actually modify files.
+        lookback_days: Only fix entries with dates within this many days. None for all.
+        github_repo: GitHub repository for PR links.
+        token: GitHub API token. If None, will be resolved from environment.
+
+    Returns:
+        List of ChangelogFixResult for each connector processed.
+    """
+    repo_path = Path(repo_path)
+    connectors = get_all_connectors(repo_path)
+
+    results = []
+    for connector_name in sorted(connectors):
+        result = fix_changelog_dates(
+            repo_path=repo_path,
+            connector_name=connector_name,
+            dry_run=dry_run,
+            lookback_days=lookback_days,
+            github_repo=github_repo,
+            token=token,
+        )
+        results.append(result)
+
+    return results