Add progress logging and reliable GitHub mirror enrichment

CHANGELOG.md

@@ -1,5 +1,39 @@
 # Changelog
 
+## v1.0.3 - 2026-10-27
+
+### Progress and diagnostics
+
+- Added configurable progress logs (PROGRESS_INTERVAL, default 60 seconds) with
+  repository, stage, pages, processed counters and elapsed time. Quiet mode keeps
+  all messages in its log. Slow operations may delay entries until a work boundary.
+- Replaced the expected nonzero worker exit for unavailable GitHub sources with
+  an explicit result marker. This removes the misleading ERR-trap Failure message
+  without suppressing genuine exit-code-4 errors such as invalid JSON.
+- The October 27 production candidate completed in 7h 49m: 59 repositories found,
+  56 successful, one unavailable source, two filtered out, zero failed. This run
+  validated the comment/redirect fixes; progress and diagnostic changes were
+  subsequently validated with mock APIs before deployment.
+
+### Fixed
+
+- Read Gitea per-issue comments once as a complete list. This endpoint does not
+  support page/limit; requesting further pages returned the same comments and
+  incorrectly stopped synchronization. Paginated repository, issue and GitHub
+  comment lists still use all pages and retain repeated-page protection.
+- Follow GitHub GET redirects (301, 302, 303, 307, 308), including root-relative
+  locations, up to five hops. Only the configured API origin is allowed, so
+  credentials cannot be forwarded to another host or a downgraded protocol.
+- Keep mutation requests non-retrying and non-redirecting. A missing GitHub
+  source (404) is counted as unavailable, preserves the mirror and does not fail
+  the run. A 404 can also mean a private source or insufficient token access.
+
+### Validation
+
+Mock API regression tests cover discovery pagination, unpaginated Gitea comments,
+repeat-run idempotency, redirects, unsafe targets, loops, rate limits, isolation
+and quiet background execution. Tests use local mock APIs, not production writes.
+
 ## v1.0.2 - 2026-10-19
 
 ### Changes
@@ -44,3 +78,7 @@ packages the scheduling setup rather than claiming a new quiet-mode code change.
   with metadata, topics, issues and comments.
 - Included dry-run, foreground apply and quiet background apply modes,
   rate-limit handling, locking and English/YunoHost documentation.
+
+GitHub sources returning HTTP 404 are counted as `unavailable`, separately from
+`failed` and discovery filters (`skipped`). Existing mirrors are preserved; the
+run can succeed with unavailable sources. Check the source or access if unexpected.

README.md

@@ -148,9 +148,43 @@ mock API tests, including multiple mirrors, idempotency, throttling and quiet mo
 
 Download the script or the complete bundle from
 [GitHub Releases](https://github.com/adamzend/gitea-github-sync/releases).
-The latest release is **v1.0.2**, with weekly root cron setup and deployment notes.
+The latest release is **v1.0.3**, adding progress logs and fixing comment retrieval, redirects and unavailable-source diagnostics.
 
 ## License
 
 This project is licensed under the GNU General Public License v3.0 (GPL-3.0-only).
 See [LICENSE](LICENSE) for the full license text.
+
+## Comment pagination and moved GitHub repositories (v1.0.3)
+
+Gitea's per-issue comments endpoint returns the complete comment list without
+pagination. The script reads it once, including lists longer than 50 comments.
+Other paginated lists are read until an empty page, with repeated-page detection.
+GitHub GET redirects are followed for up to five hops within the configured API
+origin only. External origins, protocol downgrades and redirect loops fail safely.
+HTTP 404 still requires checking the source URL and token access.
+
+After upgrading, keep the existing configuration and tokens. A failed run may
+have completed other repositories; synchronization markers make reruns idempotent.
+Quiet mode secures its log with chmod 600 on each launch, which can mask an ACL
+previously granted to a monitoring account. An administrator can restore that
+account's read ACL after launch without making the log world-readable.
+
+GitHub sources returning HTTP 404 are counted as `unavailable`, separately from
+`failed` and discovery filters (`skipped`). Existing mirrors are preserved; the
+run can succeed with unavailable sources. Check the source or access if unexpected.
+
+## Progress logging
+
+Progress is logged at the start of each stage and approximately every 60 seconds
+while processing pages, indexing issues, reading Gitea comments, mapping GitHub
+comments and comparing/applying updates. Entries include the repository, stage,
+page or processed counts and elapsed seconds since script startup. Page counts
+are completed pages; processed counts include the current item, not successful
+writes. The final summaries report completed operations.
+Set PROGRESS_INTERVAL to a positive integer in seconds (default 60, max 99999).
+Progress is checked at work boundaries: a slow API request, retry or JSON operation
+can delay the next entry. These messages are not a separate liveness watchdog.
+In --apply-quiet mode all progress goes to QUIET_LOG_FILE, not the terminal.
+An unavailable GitHub source is preserved and counted without a Failure diagnostic.
+Other failures, including JSON validation errors, still fail the repository.

README.txt

@@ -305,3 +305,37 @@ foreground synchronization engine in a detached process.
 
 API reference:
 https://docs.gitea.com/api/1.23/operations/user-list-repos/
+
+## Comment pagination and moved GitHub repositories (v1.0.3)
+
+Gitea's per-issue comments endpoint returns the complete comment list without
+pagination. The script reads it once, including lists longer than 50 comments.
+Other paginated lists are read until an empty page, with repeated-page detection.
+GitHub GET redirects are followed for up to five hops within the configured API
+origin only. External origins, protocol downgrades and redirect loops fail safely.
+HTTP 404 still requires checking the source URL and token access.
+
+After upgrading, keep the existing configuration and tokens. A failed run may
+have completed other repositories; synchronization markers make reruns idempotent.
+Quiet mode secures its log with chmod 600 on each launch, which can mask an ACL
+previously granted to a monitoring account. An administrator can restore that
+account's read ACL after launch without making the log world-readable.
+
+GitHub sources returning HTTP 404 are counted as `unavailable`, separately from
+`failed` and discovery filters (`skipped`). Existing mirrors are preserved; the
+run can succeed with unavailable sources. Check the source or access if unexpected.
+
+## Progress logging
+
+Progress is logged at the start of each stage and approximately every 60 seconds
+while processing pages, indexing issues, reading Gitea comments, mapping GitHub
+comments and comparing/applying updates. Entries include the repository, stage,
+page or processed counts and elapsed seconds since script startup. Page counts
+are completed pages; processed counts include the current item, not successful
+writes. The final summaries report completed operations.
+Set PROGRESS_INTERVAL to a positive integer in seconds (default 60, max 99999).
+Progress is checked at work boundaries: a slow API request, retry or JSON operation
+can delay the next entry. These messages are not a separate liveness watchdog.
+In --apply-quiet mode all progress goes to QUIET_LOG_FILE, not the terminal.
+An unavailable GitHub source is preserved and counted without a Failure diagnostic.
+Other failures, including JSON validation errors, still fail the repository.

gitea-github-sync.conf.example

@@ -19,3 +19,6 @@ GITHUB_RATE_LIMIT_RETRIES=2
 LOCK_FILE='/var/run/gitea-github-sync.lock'
 # Used only by --apply-quiet. Parent directory must already exist and be writable.
 QUIET_LOG_FILE='/var/log/gitea-github-sync.log'
+
+# Progress interval in seconds; slow requests may delay the next entry.
+PROGRESS_INTERVAL=60

gitea-github-sync.sh

@@ -29,6 +29,17 @@ USAGE
 }
 log() { printf '[%s] %s%s\n' "$(date '+%F %T')" "${REPO_LABEL:+[$REPO_LABEL] }" "$*" >&2; }
 die() { log "ERROR: $*"; exit 1; }
+# Throttled progress at work boundaries; no extra process or credential output.
+PROGRESS_LAST=-1
+PROGRESS_STAGE=
+progress() {
+    local stage=$1; shift
+    if [[ "$stage" != "$PROGRESS_STAGE" ]] || ((PROGRESS_LAST < 0 || SECONDS - PROGRESS_LAST >= PROGRESS_INTERVAL)); then
+        log "Progress: $stage; $*; elapsed=${SECONDS}s"
+        PROGRESS_LAST=$SECONDS
+        PROGRESS_STAGE=$stage
+    fi
+}
 on_error() { local rc=$1 line=$2; log "Failure at line $line (exit $rc); synchronization incomplete."; exit "$rc"; }
 trap 'on_error "$?" "$LINENO"' ERR
 while (($#)); do
@@ -78,6 +89,8 @@ fi
 : "${GITHUB_API_VERSION:=2026-03-10}"
 : "${LOCK_FILE:=/var/run/gitea-github-sync.lock}"
 : "${MAX_PAGES:=10000}"
+: "${PROGRESS_INTERVAL:=60}"
+[[ "$PROGRESS_INTERVAL" =~ ^[1-9][0-9]{0,4}$ ]] || die "PROGRESS_INTERVAL must be a positive integer (seconds)"
 : "${GITHUB_RATE_LIMIT_MAX_WAIT:=60}"
 : "${GITHUB_RATE_LIMIT_RETRIES:=2}"
 [[ -n "${GITEA_TOKEN:-}" ]] || die "GITEA_TOKEN is missing from the configuration"
@@ -142,7 +155,9 @@ response_header() {
 request() {
     local service=$1 method=$2 url=$3 output=$4 payload=${5:-} allow_missing=${6:-0} status
     local headers="$WORK_DIR/response.headers" message remaining reset retry_after wait_seconds
-    local attempt=0 waited=0 now reset_wait
+    local attempt=0 waited=0 now reset_wait redirects=0 location api_origin
+    api_origin=${GITHUB_API#*://}
+    api_origin="${GITHUB_API%%://*}://${api_origin%%/*}"
     local -a args=(-q --silent --show-error --connect-timeout 15 --max-time 120
         --user-agent 'gitea-github-sync/3.1' -D "$headers"
         -H "@$HEADER_DIR/$service.headers" -X "$method" -o "$output" -w '%{http_code}')
@@ -157,6 +172,18 @@ request() {
     while :; do
         status=$(curl "${args[@]}" "$url") || die "Transport error: $method $url; run dry-run before retrying APPLY"
         if [[ "$status" == 2[0-9][0-9] ]]; then return 0; fi
+        if [[ "$service" == gh && "$method" == GET && "$status" =~ ^(301|302|303|307|308)$ ]]; then
+            location=$(response_header "$headers" location)
+            # Resolve root-relative redirects; never send credentials to another origin.
+            if [[ "$location" == /* && "$location" != //* ]]; then location="$api_origin$location"; fi
+            [[ "$location" == "$api_origin/"* && "$location" != *[[:space:]]* && "$location" != *\\* ]] ||
+                die "Rejected GitHub redirect outside the configured API origin"
+            ((redirects < 5)) || die "GitHub redirect limit exceeded"
+            redirects=$((redirects + 1))
+            log "Following GitHub HTTP $status redirect ($redirects/5)"
+            url=$location
+            continue
+        fi
         if [[ "$status" == 404 && "$allow_missing" == 1 ]]; then return 4; fi
         message=$(jq -r 'if type=="object" and (.message|type)=="string" then
             .message | gsub("[[:cntrl:]]"; " ") | .[:1000]
@@ -204,13 +231,18 @@ request() {
 }
 validate() { jq -e "$2" "$1" >/dev/null || die "Invalid JSON data: ${1##*/}"; }
 paginate() {
-    local service=$1 url=$2 output=$3 page count
+    local service=$1 url=$2 output=$3 page count total=0
     : > "$output"
     for ((page=1; page<=MAX_PAGES; page++)); do
+        progress "Reading $service ${output##*/}" "page=$page, records_read=$total"
         request "$service" GET "$url&page=$page" "$WORK_DIR/page.json"
         validate "$WORK_DIR/page.json" 'type == "array" and all(.[]; type == "object")'
         count=$(jq length "$WORK_DIR/page.json")
-        ((count > 0)) || return 0
+        if ((count == 0)); then
+            log "Read complete: $service ${output##*/}; pages=$((page - 1)), records=$total"
+            return 0
+        fi
+        total=$((total + count))
         if ((page > 1)) && cmp -s "$WORK_DIR/page.json" "$WORK_DIR/previous-page.json"; then
             die "API returned a repeated page: $url"
         fi
@@ -238,6 +270,9 @@ comment_schema='def pos: type == "number" and . > 0 and floor == .;
 sync_repository() (
 set -Eeuo pipefail
 local entry=$1 repo_id source current_source
+local indexed=0 read_issues=0 read_comments=0 mapped_comments=0 processed_issues=0 processed_comments=0
+PROGRESS_LAST=-1
+PROGRESS_STAGE=
 local GITEA_OWNER GITEA_REPO GITHUB_OWNER GITHUB_REPO WORK_DIR
 repo_id=$(jq -r .id <<< "$entry")
 GITEA_OWNER=$(jq -r .owner.login <<< "$entry")
@@ -255,7 +290,17 @@ GT="$GITEA_URL/api/v1/repos/$GITEA_OWNER/$GITEA_REPO"
 GH="$GITHUB_API/repos/$GITHUB_OWNER/$GITHUB_REPO"
 log "GitHub $GITHUB_OWNER/$GITHUB_REPO -> Gitea $GITEA_OWNER/$GITEA_REPO; mode $MODE"
 request gt GET "$GT" "$WORK_DIR/gt-repo.json"
-request gh GET "$GH" "$WORK_DIR/gh-repo.json"
+if request gh GET "$GH" "$WORK_DIR/gh-repo.json" "" 1; then
+    :
+else
+    rc=$?
+    if ((rc == 4)); then
+        log "Source unavailable (GitHub HTTP 404): removed, private or inaccessible; preserving the mirror"
+        : > "$RUN_DIR/unavailable-$repo_id"
+        exit 0
+    fi
+    exit "$rc"
+fi
 for service in gt gh; do
     validate "$WORK_DIR/$service-repo.json" 'type=="object" and (.id|type=="number") and
         (.default_branch|type=="string") and (.description==null or (.description|type=="string")) and
@@ -296,6 +341,8 @@ done
 declare -A gh_numbers=() gt_numbers=() gt_comment_ids=() seen_gh=()
 while IFS= read -r record; do
     printf '%s\n' "$record" > "$WORK_DIR/record.json"
+    indexed=$((indexed + 1))
+    progress "Indexing GitHub issues" "processed=$indexed"
     id=$(jq -r .id "$WORK_DIR/record.json")
     number=$(jq -r .number "$WORK_DIR/record.json")
     [[ -z "${seen_gh[$id]+x}" && -z "${gh_numbers[$number]+x}" ]] || die "Duplicate GitHub issue $id"
@@ -304,17 +351,23 @@ while IFS= read -r record; do
 done < "$WORK_DIR/gh-issues.jsonl"
 while IFS= read -r record; do
     printf '%s\n' "$record" > "$WORK_DIR/record.json"
+    read_issues=$((read_issues + 1))
+    progress "Reading Gitea issues and comments" "issues=$read_issues, comments=$read_comments"
     id=$(marker "$WORK_DIR/record.json" github-sync-id)
     [[ -n "$id" ]] || continue
     [[ -z "${gt_numbers[$id]+x}" ]] || die "Duplicate github-sync-id:$id in Gitea"
     number=$(jq -r .number "$WORK_DIR/record.json")
     gt_numbers[$id]=$number
     printf '%s\n' "$record" > "$WORK_DIR/gt-issue-$id.json"
-    paginate gt "$GT/issues/$number/comments?limit=50" "$WORK_DIR/comments.jsonl"
+    # Gitea's per-issue comments endpoint returns the complete list and ignores
-    jq -s '.' "$WORK_DIR/comments.jsonl" > "$WORK_DIR/comments.json"
+    # page/limit. Only list endpoints that support pagination use paginate().
-    validate "$WORK_DIR/comments.json" "$comment_schema"
+    request gt GET "$GT/issues/$number/comments" "$WORK_DIR/comments.json"
+    validate "$WORK_DIR/comments.json" "type == \"array\" and $comment_schema"
+    jq -c '.[]' "$WORK_DIR/comments.json" > "$WORK_DIR/comments.jsonl"
     while IFS= read -r comment; do
         printf '%s\n' "$comment" > "$WORK_DIR/record.json"
+        read_comments=$((read_comments + 1))
+        progress "Reading Gitea issues and comments" "issues=$read_issues, current_issue=#$number, comments=$read_comments"
         cid=$(marker "$WORK_DIR/record.json" github-comment-id)
         [[ -n "$cid" ]] || continue
         key="$number:$cid"
@@ -333,6 +386,8 @@ validate "$WORK_DIR/gh-comments.json" "$comment_schema and all(.[]; (.issue_url|
 declare -A seen_comments=()
 while IFS= read -r record; do
     printf '%s\n' "$record" > "$WORK_DIR/record.json"
+    mapped_comments=$((mapped_comments + 1))
+    progress "Mapping GitHub comments" "processed=$mapped_comments"
     number=$(jq -r '.issue_url | split("/") | last' "$WORK_DIR/record.json")
     [[ "$number" =~ ^[1-9][0-9]*$ ]] || die "Invalid issue number in GitHub comment"
     [[ -n "${gh_numbers[$number]+x}" ]] || continue
@@ -381,6 +436,8 @@ if ! cmp -s "$WORK_DIR/topics-old.json" "$WORK_DIR/topics-payload.json"; then
 fi
 while IFS= read -r record; do
     printf '%s\n' "$record" > "$WORK_DIR/source.json"
+    processed_issues=$((processed_issues + 1))
+    progress "Comparing and synchronizing issues/comments" "issues=$processed_issues, comments=$processed_comments, mode=$MODE"
     id=$(jq -r .id "$WORK_DIR/source.json")
     number=$(jq -r .number "$WORK_DIR/source.json")
     target=${gt_numbers[$id]:-}
@@ -409,6 +466,8 @@ while IFS= read -r record; do
     [[ -f "$WORK_DIR/gh-comments-$number.jsonl" ]] || continue
     while IFS= read -r comment; do
         printf '%s\n' "$comment" > "$WORK_DIR/source.json"
+        processed_comments=$((processed_comments + 1))
+        progress "Comparing and synchronizing issues/comments" "issues=$processed_issues, current_issue=#$number, comments=$processed_comments, mode=$MODE"
         cid=$(jq -r .id "$WORK_DIR/source.json")
         jq "$body_filter"'{body:(.body | marked("github-comment-id"; $id))}' \
             --argjson id "$cid" "$WORK_DIR/source.json" > "$WORK_DIR/comment-payload.json"
@@ -479,7 +538,7 @@ validate "$WORK_DIR/repositories.json" 'all(.[];
     (.owner.login|type=="string" and test("^[A-Za-z0-9_.-]+$") and .!="." and .!="..") and
     (.mirror|type=="boolean") and (.original_url==null or (.original_url|type=="string")))'
 declare -A discovered_ids=()
-discovered=0 selected=0 skipped=0 succeeded=0 failed=0
+discovered=0 selected=0 skipped=0 succeeded=0 failed=0 unavailable=0
 while IFS= read -r entry; do
     repo_id=$(jq -r .id <<< "$entry")
     [[ -z "${discovered_ids[$repo_id]+x}" ]] || continue
@@ -507,9 +566,15 @@ while IFS= read -r entry; do
     sync_repository "$entry" &
     worker_pid=$!
     if wait "$worker_pid"; then
-        succeeded=$((succeeded + 1))
+        if [[ -f "$RUN_DIR/unavailable-$repo_id" ]]; then
-        log "OK $full_name"
+            unavailable=$((unavailable + 1))
+            log "UNAVAILABLE $full_name: mirror preserved"
+        else
+            succeeded=$((succeeded + 1))
+            log "OK $full_name"
+        fi
     else
+        rc=$?
         failed=$((failed + 1))
         log "FAILED $full_name"
     fi
@@ -519,6 +584,6 @@ while IFS= read -r entry; do
         break
     fi
 done < "$WORK_DIR/repositories.jsonl"
-log "Repositories: discovered=$discovered, selected=$selected, succeeded=$succeeded, failed=$failed, skipped=$skipped"
+log "Repositories: discovered=$discovered, selected=$selected, succeeded=$succeeded, failed=$failed, skipped=$skipped, unavailable=$unavailable"
 ((failed == 0)) || exit 1
 log "Account synchronization completed ($MODE)"

tests/test_error_status.py

@@ -0,0 +1,30 @@
+import subprocess
+import tempfile
+import threading
+from pathlib import Path
+from http.server import ThreadingHTTPServer
+
+# Reuse the mock API definitions without running its full regression suite.
+ns = {'__file__': str(Path(__file__).with_name('test_sync.py'))}
+exec(Path(ns['__file__']).read_text().split('server=ThreadingHTTPServer')[0], ns)
+class InvalidJSON(ns['Handler']):
+    def do_GET(self):
+        if self.path == '/gh/repos/first/alpha':
+            self.send_response(200)
+            self.end_headers()
+            self.wfile.write(b'not-json')
+        else:
+            super().do_GET()
+ns['reset']()
+ns['repos'] = ns['repos'][:1]
+server = ThreadingHTTPServer(('127.0.0.1', 0), InvalidJSON)
+threading.Thread(target=server.serve_forever, daemon=True).start()
+with tempfile.TemporaryDirectory() as tmp:
+    config = Path(tmp) / 'config'
+    config.write_text(f"GITEA_TOKEN=test\nGITEA_URL=http://127.0.0.1:{server.server_port}\nGITHUB_API=http://127.0.0.1:{server.server_port}/gh\nLOCK_FILE={tmp}/lock\n")
+    result = subprocess.run(['bash', ns['SCRIPT'], '--config', str(config), '--apply', '--no-git'], capture_output=True, text=True)
+    assert result.returncode != 0, result.stderr
+    assert 'failed=1' in result.stderr and 'unavailable=0' in result.stderr, result.stderr
+    assert 'Invalid JSON' in result.stderr and not ns['writes'], result.stderr
+server.shutdown()
+print('PASS invalid JSON remains a genuine failure, not an unavailable source')

tests/test_progress.py

@@ -0,0 +1,20 @@
+import subprocess
+from pathlib import Path
+
+source = (Path(__file__).resolve().parents[1] / 'gitea-github-sync.sh').read_text()
+progress = source[source.index('PROGRESS_LAST=-1'):source.index('on_error() {')]
+program = "set -Eeuo pipefail\nlog() { printf '%s\\n' \"$*\"; }\nPROGRESS_INTERVAL=60\n" + progress + """
+SECONDS=0
+progress "Reading comments" "processed=0"
+SECONDS=59
+progress "Reading comments" "processed=59"
+SECONDS=60
+progress "Reading comments" "processed=60"
+progress "Applying issues" "processed=0"
+"""
+result = subprocess.run(['bash', '-c', program], capture_output=True, text=True, check=True)
+lines = result.stdout.splitlines()
+assert len(lines) == 3, result.stdout
+assert 'processed=59' not in result.stdout
+assert 'processed=60' in lines[1] and 'Applying issues' in lines[2]
+print('PASS progress interval, counters and immediate stage transitions')

tests/test_sync.py

@@ -0,0 +1,239 @@
+import copy
+import json
+import subprocess
+import tempfile
+import threading
+import time
+from pathlib import Path
+from http.server import BaseHTTPRequestHandler, ThreadingHTTPServer
+from urllib.parse import urlsplit, parse_qs
+
+SCRIPT = str(Path(__file__).resolve().parents[1] / 'gitea-github-sync.sh')
+repos = []
+reads = []
+writes = []
+fail_repo = None
+bad_discovery = False
+failure_mode = 'permission'
+request_gate = None
+redirect_mode = None
+
+def make(n, source, **kwargs):
+    r = dict(id=n, name=f'local-{n}', owner={'login':'adam'}, mirror=True,
+             original_url=source, archived=False, has_issues=True,
+             default_branch='main', description='old', website='', topics=[], issues=[], comments=[])
+    r.update(kwargs)
+    return r
+
+def reset():
+    global repos, fail_repo, bad_discovery
+    fail_repo = None
+    bad_discovery = False
+    reads.clear()
+    writes.clear()
+    repos = [make(1, 'https://github.com/first/alpha.git'),
+             make(2, 'git@github.com:second/beta.git', private=True),
+             make(3, 'ssh://git@github.com:22/third/gamma.git/', has_issues=False),
+             make(4, 'https://gitlab.com/other/project.git'),
+             make(5, 'https://github.com/first/alpha.git', mirror=False),
+             make(6, 'https://github.com/first/alpha.git', owner={'login':'someone-else'}),
+             make(7, 'https://github.com.evil.invalid/a/b.git'),
+             make(8, 'https://user:password@GitHub.com/Fourth/delta.git/')]
+
+class Handler(BaseHTTPRequestHandler):
+    def log_message(self, *args): pass
+    def answer(self, code, obj=None, headers=None):
+        self.send_response(code)
+        self.send_header('Content-Type', 'application/json')
+        for k,v in (headers or {}).items(): self.send_header(k,str(v))
+        self.end_headers()
+        if obj is not None: self.wfile.write(json.dumps(obj).encode())
+    def do_GET(self):
+        global fail_repo
+        if request_gate is not None: request_gate.wait(timeout=10)
+        reads.append(self.path)
+        u = urlsplit(self.path)
+        page = int(parse_qs(u.query).get('page', ['1'])[0])
+        if u.path == '/api/v1/user': return self.answer(200, {'login':'adam'})
+        if u.path == '/api/v1/users/adam/repos':
+            return self.answer(200, {'error':'invalid'} if bad_discovery else repos[page-1:page])
+        if u.path.startswith('/gh/') and redirect_mode:
+            if redirect_mode == 'external':
+                return self.answer(301, {}, {'Location':'https://example.invalid/steal'})
+            if redirect_mode == 'loop':
+                return self.answer(301, {}, {'Location':self.path})
+            if not u.path.startswith('/gh/moved/'):
+                return self.answer(301, {}, {'Location':self.path.replace('/gh/', '/gh/moved/', 1)})
+        gh = u.path.startswith('/gh/')
+        parts = u.path.split('/repos/')[-1].split('/')
+        owner, name = parts[:2]
+        tail = '/' + '/'.join(parts[2:]) if len(parts)>2 else ''
+        mapping = {'alpha':1, 'beta':2, 'gamma':3, 'delta':8}
+        if gh:
+            if name not in mapping: return self.answer(404, {})
+            n = mapping[name]
+            expected = {1:'first',2:'second',3:'third',8:'Fourth'}[n]
+            assert owner == expected
+        else:
+            assert owner == 'adam'
+            n = int(name.removeprefix('local-'))
+        r = next(x for x in repos if x['id']==n)
+        if gh and n == fail_repo:
+            if failure_mode == 'primary':
+                return self.answer(403, {'message':'API rate limit exceeded'}, {'X-RateLimit-Remaining':0,'X-RateLimit-Reset':int(time.time())+3600})
+            if failure_mode == 'secondary':
+                return self.answer(403, {'message':'You have exceeded a secondary rate limit'})
+            if failure_mode == 'retry':
+                fail_repo = None
+                return self.answer(429, {'message':'Too many requests'}, {'Retry-After':1})
+            if failure_mode == 'missing': return self.answer(404, {'message':'Not Found'})
+            if failure_mode == 'auth': return self.answer(401, {'message':'Bad credentials'})
+            return self.answer(403, {'message':'denied'})
+        if tail == '': return self.answer(200, dict(id=n, default_branch='main', description=f'source-{n}', homepage=f'https://example.com/{n}') if gh else r)
+        if tail == '/topics': return self.answer(200, {'names' if gh else 'topics':[f'topic-{n}'] if gh else r['topics']})
+        if tail == '/issues':
+            assert r['has_issues'], 'must not request disabled issues'
+            data = [dict(id=100+n, number=1, title=f'Issue-{n}', body=f'Body-{n}', state='closed')] if gh else r['issues']
+        elif tail == '/issues/comments':
+            assert r['has_issues']
+            data = [dict(id=200+n, body=f'Comment-{n}', issue_url=f'https://api.github.com/repos/{owner}/{name}/issues/1')] if gh else r['comments']
+        elif tail.endswith('/comments'): return self.answer(200, r['comments'])
+        else: return self.answer(404, {})
+        self.answer(200, data[page-1:page])
+    def write(self):
+        data = json.loads(self.rfile.read(int(self.headers.get('Content-Length',0))) or '{}')
+        parts = self.path.split('/repos/adam/local-')[1].split('/')
+        n = int(parts[0]); tail = '/'+'/'.join(parts[1:]) if len(parts)>1 else ''
+        writes.append((n,self.command,tail,data))
+        r = next(x for x in repos if x['id']==n)
+        if tail == '': r.update(data); return self.answer(200,r)
+        if tail == '/topics': r['topics']=data['topics']; return self.answer(204)
+        if tail == '/mirror-sync': return self.answer(200)
+        if tail == '/issues':
+            item = dict(data,id=1000+n,number=10,state='closed' if data['closed'] else 'open')
+            r['issues'].append(item); return self.answer(201,item)
+        if tail.endswith('/comments'):
+            item = dict(data,id=2000+n); r['comments'].append(item); return self.answer(201,item)
+        self.answer(404,{})
+    do_POST=write
+    do_PATCH=write
+    do_PUT=write
+
+server=ThreadingHTTPServer(('127.0.0.1',0),Handler)
+threading.Thread(target=server.serve_forever,daemon=True).start()
+with tempfile.TemporaryDirectory() as tmp:
+    conf=Path(tmp)/'config'
+    conf.write_text(f"GITEA_TOKEN='test'\nGITEA_URL='http://127.0.0.1:{server.server_port}'\nGITHUB_API='http://127.0.0.1:{server.server_port}/gh'\nLOCK_FILE='{tmp}/lock'\nGITEA_REPO='obsolete-name'\nGITHUB_RATE_LIMIT_MAX_WAIT=2\n")
+    def run(mode='--apply', success=True):
+        p=subprocess.run(['bash',SCRIPT,'--config',str(conf),mode,'--no-git'],capture_output=True,text=True)
+        assert (p.returncode==0)==success,p.stderr
+        assert 'password' not in p.stderr
+        return p.stderr
+    reset()
+    out=run('--dry-run')
+    assert not writes and 'selected=4, succeeded=4, failed=0, skipped=4' in out,out
+    assert any('page=9' in p for p in reads)
+    assert 'Progress: Reading gt repositories.jsonl; page=1' in out,out
+    assert 'Progress: Comparing and synchronizing issues/comments' in out,out
+    print('PASS account discovery, full pagination, private mirror and source URL filters')
+    run()
+    assert {n for n,_,_,_ in writes}=={1,2,3,8}
+    for r in repos:
+        if r['id'] in {1,2,8}:
+            assert len(r['issues'])==len(r['comments'])==1
+            assert r['issues'][0]['title']==f"Issue-{r['id']}"
+            assert r['comments'][0]['body'].startswith(f"Comment-{r['id']}")
+        if r['id']==3: assert not r['issues'] and r['description']=='source-3'
+    writes.clear()
+    run()
+    assert not writes
+    print('PASS isolated state, renamed destinations, disabled issues, multi-repository idempotency')
+    repos[0]['comments'].extend(dict(id=9000+i,body=f'Local comment {i}') for i in range(60))
+    reads.clear(); run(); assert not writes
+    assert len([p for p in reads if p == '/api/v1/repos/adam/local-1/issues/10/comments']) == 1
+    assert len(repos[0]['comments']) == 61
+    print('PASS complete unpaginated list over 50 comments read once without duplicate writes')
+    reset(); fail_repo=1
+    out=run(success=False)
+    assert {n for n,_,_,_ in writes}=={2,3,8}
+    assert 'selected=4, succeeded=3, failed=1' in out,out
+    print('PASS one repository failure does not stop remaining repositories')
+    reset(); fail_repo=1; failure_mode='missing'
+    out=run()
+    assert 'unavailable=1' in out and 'failed=0' in out,out
+    assert 'Failure at line' not in out and 'FAILED ' not in out,out
+    assert {n for n,_,_,_ in writes}=={2,3,8}
+    failure_mode='permission'
+    print('PASS missing GitHub source preserved and counted separately without failing the run')
+    reset(); repos[0]['archived']=True
+    out=run(success=False)
+    assert {n for n,_,_,_ in writes}=={2,3,8}
+    print('PASS archived mirrors reported without blocking other repositories')
+    reset(); repos=[]
+    out=run()
+    assert not writes and 'discovered=0' in out
+    print('PASS empty account')
+    reset(); bad_discovery=True
+    run(success=False)
+    assert not writes
+    print('PASS discovery failure prevents writes')
+    for failure_mode in ['primary','secondary','auth']:
+        reset(); fail_repo=1
+        out=run(success=False)
+        assert not writes and 'Account scan aborted' in out,out
+        assert len([p for p in reads if p.startswith('/gh/')])==1,reads
+        if failure_mode=='primary': assert 'quota resets at' in out
+        if failure_mode=='auth': assert 'Bad credentials' in out
+        print(f'PASS {failure_mode} error is diagnosed and stops further GitHub requests')
+    failure_mode='retry'
+    reset(); fail_repo=1
+    out=run()
+    assert 'Waiting 1 seconds' in out and 'succeeded=4' in out,out
+    assert {n for n,_,_,_ in writes}=={1,2,3,8}
+    print('PASS Retry-After respected and synchronization resumes')
+    reset(); redirect_mode='same'
+    out=run()
+    assert 'Following GitHub HTTP 301' in out and 'succeeded=4' in out,out
+    writes.clear(); run(); assert not writes
+    print('PASS same-origin redirects and idempotent unpaginated Gitea comments')
+    for redirect_mode in ['external','loop']:
+        reset(); out=run(success=False)
+        assert not writes
+        assert ('Rejected GitHub redirect' if redirect_mode=='external' else 'redirect limit exceeded') in out,out
+    redirect_mode=None
+    print('PASS cross-origin redirects and redirect loops rejected before writes')
+    reset()
+    quiet_log=Path(tmp)/'quiet.log'
+    with conf.open('a') as f: f.write(f"QUIET_LOG_FILE='{quiet_log}'\n")
+    request_gate=threading.Event()
+    p=subprocess.run(['bash',SCRIPT,'--config',str(conf),'--apply-quiet','--no-git'],capture_output=True,text=True,timeout=5)
+    assert p.returncode==0 and not p.stdout and not p.stderr,p
+    assert not writes, 'launcher must return before blocked background work completes'
+    request_gate.set()
+    request_gate=None
+    deadline=time.monotonic()+60
+    while time.monotonic()<deadline:
+        if quiet_log.exists() and 'Background worker finished:' in quiet_log.read_text(): break
+        time.sleep(.1)
+    out=quiet_log.read_text()
+    assert 'exit_status=0' in out and 'succeeded=4' in out,out
+    assert 'Progress:' in out,out
+    assert {n for n,_,_,_ in writes}=={1,2,3,8}
+    assert quiet_log.stat().st_mode & 0o777 == 0o600
+    print('PASS quiet mode returns immediately, writes no terminal output, applies all repositories and logs completion')
+    reset(); failure_mode='auth'; fail_repo=1
+    quiet_log.write_text('')
+    p=subprocess.run(['bash',SCRIPT,'--config',str(conf),'--apply-quiet'],capture_output=True,text=True,timeout=5)
+    assert p.returncode==0 and not p.stdout and not p.stderr
+    deadline=time.monotonic()+30
+    while time.monotonic()<deadline:
+        if 'Background worker finished:' in quiet_log.read_text(): break
+        time.sleep(.1)
+    assert 'exit_status=1' in quiet_log.read_text() and not writes
+    print('PASS quiet background failure is retained in the log')
+    badconf=Path(tmp)/'bad-config'
+    badconf.write_text(conf.read_text()+f"QUIET_LOG_FILE='{tmp}/missing/log'\n")
+    p=subprocess.run(['bash',SCRIPT,'--config',str(badconf),'--apply-quiet'],capture_output=True,text=True)
+    assert p.returncode!=0 and not p.stdout and not p.stderr
+    print('PASS inaccessible quiet log fails silently before launching')
+server.shutdown()