feat: Markdown emphasis inside columns blocks

Inside a columns fenced block (verbatim, so Pandoc doesn't process
Markdown), convert *italic* / **bold** / ***bold-italic*** to wikitext
emphasis. Asterisk style, single line; precedence bold-italic > bold >
italic. Raw inline HTML still works via the existing HTML-unescape path.

Docs + example updated to use ** ** rather than <b>.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

.gitea/workflows/build-image.yml

@@ -0,0 +1,99 @@
+on:
+  push:
+    branches: [master]
+    paths:
+      - Dockerfile
+      - "pipelines/**/requirements.txt"
+      - .gitea/workflows/build-image.yml
+  workflow_dispatch:
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+    container:
+      # Daemonless image builder: kaniko :debug plus the /bin/sleep symlink
+      # act needs for job-container PID 1 (see ci/Dockerfile.kaniko-act).
+      # No docker socket, no privileges.
+      image: pi5-16.local:3005/mikkeli/kaniko-act:v1.23.2-r2
+
+    steps:
+      - name: Fetch source, build, push
+        # kaniko image has no bash — busybox sh only
+        shell: sh
+        env:
+          # package r/w PAT (user-level secret) — registry push only
+          PKG_TOKEN: ${{ secrets.PKGRW_PAT }}
+          # per-run auto-token, read access to this repo only — source fetch
+          JOB_TOKEN: ${{ github.token }}
+          URL_TO_GITEA: ${{ vars.URL_TO_GITEA }}
+        run: |
+          set -eu
+          HOST="${URL_TO_GITEA#http://}"
+          SHORT_SHA=$(echo "${GITHUB_SHA}" | cut -c1-7)
+          IMAGE="${HOST}/mikkeli/novoyuuparosk-wiki-runner"
+
+          # Source via Gitea archive API (no git/node in this image; repo is
+          # private — authenticate with the job's own per-run token)
+          wget -qO /tmp/src.tar.gz \
+            "http://mikkeli:${JOB_TOKEN}@${HOST}/api/v1/repos/mikkeli/novoyuuparosk-auto-wiki/archive/${GITHUB_SHA}.tar.gz"
+          mkdir -p /tmp/src
+          tar -xzf /tmp/src.tar.gz -C /tmp/src --strip-components=1
+
+          # Registry auth for the push
+          AUTH=$(printf '%s' "mikkeli:${PKG_TOKEN}" | base64 | tr -d '\n')
+          printf '{"auths":{"%s":{"auth":"%s"}}}' "${HOST}" "${AUTH}" \
+            > /kaniko/.docker/config.json
+
+          # --insecure: push target is the plain-HTTP Gitea registry.
+          # Base image pull (docker.io) stays HTTPS — no --insecure-pull.
+          # :latest is a convenience tag for internal CI (the pin job below
+          # runs in it). Publish workflows stay pinned to the immutable sha.
+          /kaniko/executor \
+            --context dir:///tmp/src \
+            --dockerfile /tmp/src/Dockerfile \
+            --destination "${IMAGE}:${SHORT_SHA}" \
+            --destination "${IMAGE}:latest" \
+            --insecure
+
+          echo "Pushed ${IMAGE}:${SHORT_SHA} and ${IMAGE}:latest"
+
+  pin:
+    needs: build
+    runs-on: ubuntu-latest
+    container:
+      # Freshly built runner image: git + GNU sed baked in, and it has no
+      # non-shell ENTRYPOINT, so it works as a job container as-is (unlike the
+      # kaniko image above). Pulls :latest that the build job just pushed.
+      image: pi5-16.local:3005/mikkeli/novoyuuparosk-wiki-runner:latest
+    steps:
+      - name: Repoint publish workflows at the new image tag
+        env:
+          # Full-access PAT — needs contents:write to push the pin commit.
+          # github.token is read-only here; PKGRW_PAT is registry-only.
+          FAPAT: ${{ secrets.FAPAT }}
+          URL_TO_GITEA: ${{ vars.URL_TO_GITEA }}
+        run: |
+          set -eu
+          HOST="${URL_TO_GITEA#http://}"
+          SHORT_SHA=$(echo "${GITHUB_SHA}" | cut -c1-7)
+
+          git clone "http://mikkeli:${FAPAT}@${HOST}/mikkeli/novoyuuparosk-auto-wiki" repo
+          cd repo
+
+          for f in .gitea/workflows/publish-songs.yml \
+                   .gitea/workflows/publish-ses.yml \
+                   .gitea/workflows/publish-tech.yml; do
+            sed -i -E "s#(novoyuuparosk-wiki-runner:)[A-Za-z0-9._-]+#\1${SHORT_SHA}#" "$f"
+          done
+
+          if git diff --quiet; then
+            echo "Pins already at ${SHORT_SHA}; nothing to commit."
+            exit 0
+          fi
+
+          git config user.email "actions@novoyuuparosk.org"
+          git config user.name "wiki-runner image bot"
+          git commit -am "build: pin runner image to ${SHORT_SHA}"
+          # Pin commit only touches publish-*.yml (on: workflow_call) — triggers
+          # nothing, so no rebuild loop. Fails loudly if master moved meanwhile.
+          git push origin HEAD:master

.gitea/workflows/publish-ses.yml

@@ -10,7 +10,7 @@ jobs:
   publish:
     runs-on: ubuntu-latest
     container:
-      image: novoyuuparosk-wiki-runner:latest
+      image: pi5-16.local:3005/mikkeli/novoyuuparosk-wiki-runner:b2714ab
 
     steps:
       - name: Checkout ses-light-novel

.gitea/workflows/publish-songs.yml

@@ -19,7 +19,7 @@ jobs:
   publish:
     runs-on: ubuntu-latest
     container:
-      image: novoyuuparosk-wiki-runner:latest
+      image: pi5-16.local:3005/mikkeli/novoyuuparosk-wiki-runner:b2714ab
 
     steps:
       - name: Checkout ncmr-songs

.gitea/workflows/publish-tech.yml

@@ -10,7 +10,7 @@ jobs:
   publish:
     runs-on: ubuntu-latest
     container:
-      image: novoyuuparosk-wiki-runner:latest
+      image: pi5-16.local:3005/mikkeli/novoyuuparosk-wiki-runner:b2714ab
 
     steps:
       - name: Checkout tech-blogs

.gitignore

@@ -1,2 +1,5 @@
 # claude local settings
-.claude/
+.claude/
+
+# python
+__pycache__/

README.md

@@ -8,7 +8,7 @@ CI/CD pipelines that auto-apply commits to https://wiki.novoyuuparosk.org from u
 |---|---|---|---|
 | [`pipelines/songs/`](pipelines/songs/) | `mikkeli/ncmr-songs` | Song lyric pages | v1 live |
 | [`pipelines/ses/`](pipelines/ses/) | `mikkeli/ses-light-novel` | SES light novel pages | v1 in development |
-| [`pipelines/tech/`](pipelines/tech/) | `mikkeli/tech-blogs` | Tech blog posts | v1 ready (source repo not yet initialised) |
+| [`pipelines/tech/`](pipelines/tech/) | `mikkeli/tech-blogs` | Tech blog posts | v1 live |
 
 Per-pipeline READMEs cover everything specific to that pipeline (source schema, renderer, runtime, decisions). This root README covers only what's cross-cutting.
 
@@ -49,14 +49,11 @@ Container network mode: `host` — required so job containers can reach `localho
 
 ### Job container image
 
-All pipelines share a single pre-built Docker image: `novoyuuparosk-wiki-runner:latest`. The `Dockerfile` is at the repo root. It bakes in system deps (git, pandoc, ca-certificates) and all pipeline Python packages so job containers start instantly with no install steps.
+All pipelines share a single pre-built Docker image, served from the Gitea registry at `pi5-16.local:3005/mikkeli/novoyuuparosk-wiki-runner`. The `Dockerfile` is at the repo root. It bakes in system deps (git, pandoc, ca-certificates) and all pipeline Python packages so job containers start instantly with no install steps.
 
-The image is built manually on the Pi and stored in the local Docker daemon (`pull_image: false` in act_runner config). Rebuild after any change to the `Dockerfile` or a pipeline `requirements.txt`:
+The image builds automatically via [`.gitea/workflows/build-image.yml`](.gitea/workflows/build-image.yml), which triggers on pushes that touch the `Dockerfile`, any pipeline `requirements.txt`, or that workflow itself. It uses kaniko (daemonless, unprivileged) to build and push two tags: an immutable `:<short-sha>` and a moving `:latest`.
 
-```bash
+A follow-up `pin` job then rewrites the `image:` pin in each `publish-*.yml` to the new `:<short-sha>` and commits it back to `master` (using `FAPAT` for contents write). The publish workflows therefore always reference an immutable tag, kept current automatically — no manual bump. The pin commit only touches `workflow_call` files, so it triggers no further runs.
-docker build -t novoyuuparosk-wiki-runner:latest \
-  /home/mikkeli/dev/novoyuuparosk-auto-wiki
-```
 
 ## Gitea Actions setup (cross-cutting)
 
@@ -89,6 +86,7 @@ Secrets and variables are scoped to user `mikkeli` (no orgs on this instance), i
 | MediaWiki API path | `api.php` (classic action API) | 2026-06-09 |
 | Branch naming (this repo) | `automation/<pipeline>` for pipeline-development branches | 2026-06-09 |
 | Variable naming | `URL_TO_GITEA` not `GITEA_URL` — Gitea blocks `GITEA_`/`GITHUB_` prefixes | 2026-06-09 |
+| Columns shorthand | Side-by-side columns authored as a ` ```columns ` fenced block, expanded post-Pandoc in shared `lib/wiki.py` (universal across pipelines). No wiki template or PHP extension — runs Pi-side before the API call | 2026-06-14 |
 
 Per-pipeline decisions live in each pipeline's README.
 

ci/Dockerfile.kaniko-act

@@ -0,0 +1,14 @@
+# Kaniko executor adapted for Gitea act_runner job containers.
+#
+# act starts job containers with entrypoint /bin/sleep and sets
+# RUNNER_TEMP=/tmp; kaniko's scratch-based image has neither /bin/sleep
+# (busybox lives under /busybox; /bin/sh is already symlinked) nor /tmp.
+# This wrapper adds exactly those two. Nothing else changes.
+#
+# Bootstrap: the first build of this image is done manually on the host
+# (docker build -f ci/Dockerfile.kaniko-act -t <registry>/mikkeli/kaniko-act:<ver> .)
+# because no builder image exists yet. Later version bumps can be built by
+# the build-image workflow itself, using the previous kaniko-act.
+FROM gcr.io/kaniko-project/executor:v1.23.2-debug
+SHELL ["/busybox/sh", "-c"]
+RUN ln -sf /busybox/sleep /bin/sleep && mkdir -p -m 1777 /tmp

lib/wiki.py

@@ -1,6 +1,8 @@
 """Shared utilities for novoyuuparosk-auto-wiki pipelines."""
 
+import html
 import os
+import re
 import subprocess
 from urllib.parse import urlparse
 
@@ -8,6 +10,25 @@ import mwclient
 
 AUTO_BANNER_PREFIX = "{{Auto-generated"
 
+# A ```columns fenced block is passed through Pandoc verbatim as
+# <pre class="columns">…</pre>; columns within it are separated by a line of ===.
+_COLUMNS_BLOCK_RE = re.compile(r'<pre class="columns">(.*?)</pre>', re.DOTALL)
+_COLUMN_SEP_RE = re.compile(r"^\s*===\s*$", re.MULTILINE)
+
+# Minimal Markdown emphasis → wikitext, for use inside a verbatim columns block
+# (Pandoc doesn't process Markdown there). Asterisk style only, single line.
+# Order matters: bold-italic (***) before bold (**) before italic (*).
+_BOLD_ITALIC_RE = re.compile(r"\*\*\*(.+?)\*\*\*")
+_BOLD_RE = re.compile(r"\*\*(.+?)\*\*")
+_ITALIC_RE = re.compile(r"\*(.+?)\*")
+
+
+def _md_emphasis_to_wikitext(text: str) -> str:
+    text = _BOLD_ITALIC_RE.sub(r"'''''\1'''''", text)
+    text = _BOLD_RE.sub(r"'''\1'''", text)
+    text = _ITALIC_RE.sub(r"''\1''", text)
+    return text
+
 
 def strip_first_h1(text: str) -> str:
     """Remove the first '# Heading' line and any immediately following blank line."""
@@ -21,6 +42,29 @@ def strip_first_h1(text: str) -> str:
     return "\n".join(lines)
 
 
+def expand_columns(wikitext: str) -> str:
+    """Expand ```columns fenced blocks into a flex row of <poem> columns.
+
+    Authors write a fenced code block tagged ``columns``; Pandoc passes its body
+    through verbatim as ``<pre class="columns">…</pre>`` (line breaks and blank
+    lines preserved, inline markup entity-escaped). Columns within the block are
+    separated by a line containing only ``===``. Each column is wrapped in
+    <poem> so its line breaks survive MediaWiki parsing. Content is
+    HTML-unescaped (so inline markup such as <b>…</b> renders rather than
+    appearing as literal text) and Markdown emphasis (``*italic*``, ``**bold**``,
+    ``***bold-italic***``) is converted to wikitext, since Pandoc does not
+    process Markdown inside the verbatim block.
+    """
+
+    def render(match: re.Match) -> str:
+        body = _md_emphasis_to_wikitext(html.unescape(match.group(1)))
+        columns = _COLUMN_SEP_RE.split(body)
+        poems = "".join("<poem>\n" + col.strip("\n") + "\n</poem>" for col in columns)
+        return '<div style="display:flex; gap:3em; align-items:flex-start">' + poems + "</div>"
+
+    return _COLUMNS_BLOCK_RE.sub(render, wikitext)
+
+
 def markdown_to_wikitext(body: str) -> str:
     result = subprocess.run(
         ["pandoc", "-f", "markdown", "-t", "mediawiki"],
@@ -30,7 +74,7 @@ def markdown_to_wikitext(body: str) -> str:
     )
     if result.returncode != 0:
         raise RuntimeError(f"pandoc failed: {result.stderr.strip()}")
-    return result.stdout
+    return expand_columns(result.stdout)
 
 
 def connect_wiki() -> mwclient.Site:

pipelines/songs/README.md

@@ -89,6 +89,7 @@ The pipeline always runs `--all`: every file with `wiki.publish: true` (not unde
 | Publish mode | Always `--all`; no diff detection — small repo, simpler than fragile git-diff gating | 2026-06-09 |
 | Manual-page protection | Bot skips pages without the `{{Auto-generated` banner to avoid overwriting hand-written content | 2026-06-09 |
 | `workflow_call` abandoned | Cross-repo `workflow_call` fails — run token scoped to triggering repo; cannot clone private callee | 2026-06-09 |
+| Side-by-side columns | Authored as a ` ```columns ` fenced block (separator `===`); Pandoc emits `<pre class="columns">`, expanded post-Pandoc into a flex `<div>` of `<poem>` columns. No wiki template or extension — all Pi-side before the API write. See [SCHEMA.md](SCHEMA.md#columns-side-by-side-shorthand) | 2026-06-14 |
 
 ## Status
 

pipelines/songs/SCHEMA.md

@@ -122,20 +122,46 @@ The siblings list is a bill of materials — declaring a sibling has no effect u
 
 ## Body
 
-Plain markdown. The renderer applies two transformations before and after Pandoc:
+Plain markdown. The renderer applies these transformations before and after Pandoc:
 
 1. **Pre-Pandoc**: strip the first-line `# Heading` if present.
-2. **Post-Pandoc**: substitute `TAG` placeholders declared in `wiki.siblings`.
+2. **Post-Pandoc**: expand `columns` fenced blocks into side-by-side wikitext (see below).
+3. **Post-Pandoc**: substitute `TAG` placeholders declared in `wiki.siblings`.
 
 Do not embed raw wikitext-specific syntax (`{{Template}}`, raw `[[Wikilink]]` not declared via `siblings`, etc.) in the body unless you intend the literal output. Use markdown idioms; the renderer adds the metadata-derived bits (banner, LRC link, categories, sibling resolution) around Pandoc's output.
 
+### Columns (side-by-side) shorthand
+
+For parallel content — e.g. an original and its translation — write a fenced code block tagged `columns` and separate the columns with a line containing only `===`:
+
+````markdown
+```columns
+**原題**
+
+一行目
+二行目
+===
+**Title**
+
+first line
+second line
+```
+````
+
+The renderer turns this into a flexbox row of `<poem>` columns (one per `===`-delimited section). Notes:
+
+- **Line breaks and blank lines are preserved verbatim** — that's the point of using a fenced block; Pandoc passes the body through untouched, and each column is wrapped in `<poem>` so MediaWiki keeps the line breaks.
+- **N columns**: use N−1 `===` separators. Two is the common case (original / translation).
+- **Emphasis in Markdown**: write `*italic*`, `**bold**`, or `***bold-italic***` (asterisk style, single line) — these are converted to wikitext. Raw inline HTML such as `<b>…</b>` or `<br>` also still works (it's HTML-unescaped on the way out), but you shouldn't need it.
+- This is a shared transform (`lib/wiki.py`), so it works for any pipeline, not just songs. Column width/gap styling currently lives in that transform.
+
 ## Renderer behaviour
 
 For each `.md` with `wiki.publish: true` and not under `wip/`:
 
 1. Parse and validate frontmatter.
 2. Strip the leading `# Heading` from the body if present.
-3. Pipe the body through `pandoc -f markdown -t mediawiki`.
+3. Pipe the body through `pandoc -f markdown -t mediawiki`, then expand any `columns` fenced blocks into side-by-side `<poem>` columns.
 4. Substitute `TAG` placeholders with resolved `[[Page|Display]]` wikilinks (or fallback text).
 5. Prepend the auto-generated banner: `{{Auto-generated|source=<source URL>|commit=<sha>}}`.
 6. If `lrc` is declared: upload the LRC file if its SHA1 has changed; append a `[[Media:…]]` link.

pipelines/tech/README.md

@@ -10,7 +10,7 @@ Publishes tech blog posts from `mikkeli/tech-blogs` to `https://wiki.novoyuuparo
 - YAML frontmatter parsing per [SCHEMA.md](SCHEMA.md)
 - Pandoc-based markdown → wikitext body rendering (h1 elements preserved)
 - Auto-generated banner (`{{Auto-generated|source=...|commit=<sha>}}`)
-- Category injection: `[[Category:Tech blog]]` on every page; `[[Category:Blog:<year>]]` from `date` field; `[[Category:<tag>]]` per tag entry
+- Category injection: `[[Category:Tech blog]]` on every page; `[[Category:Blog:<year>]]` from `date` field (falling back to the action's execution year); `[[Category:<tag>]]` per tag entry
 - No title prefix — pages live at their bare title
 - MediaWiki bot API write with idempotency — no-op if wiki content matches generated output
 - Files without `wiki.publish: true` silently skipped
@@ -65,7 +65,7 @@ Shared code lives in `lib/wiki.py` (repo root).
 |---|---|---|
 | No title prefix | Tech posts live at bare titles — no namespace needed | 2026-06-10 |
 | Fixed category | Every post gets `Category:Tech blog` | 2026-06-10 |
-| Year category | `Category:Blog:<year>` from `date` field; omitted if no date | 2026-06-10 |
+| Year category | `Category:Blog:<year>` from `date` field; falls back to the execution year if no date | 2026-06-10 |
 | Tag categories | Each tag → `[[Category:<tag>]]` with no prefix | 2026-06-10 |
 | WIP exclusion | None — `wiki.publish: false` is the only gate | 2026-06-10 |
 | Publish mode | Always `--all` | 2026-06-10 |

pipelines/tech/SCHEMA.md

@@ -45,7 +45,7 @@ wiki:
 
 #### `date` (ISO 8601 date)
 
-Publication date in `YYYY-MM-DD` format. Drives `[[Category:Blog:<year>]]` injection. If omitted, no year category is added.
+Publication date in `YYYY-MM-DD` format. Drives `[[Category:Blog:<year>]]` injection. If omitted, the year falls back to the time the publish action runs.
 
 #### `tags` (list of strings)
 
@@ -69,7 +69,7 @@ For each `.md` with `wiki.publish: true`:
 1. Parse and validate frontmatter.
 2. Pipe the body through `pandoc -f markdown -t mediawiki`.
 4. Prepend the auto-generated banner: `{{Auto-generated|source=<source URL>|commit=<sha>}}`.
-5. Append category tags: `[[Category:Tech blog]]`; `[[Category:Blog:<year>]]` if `date` is set; `[[Category:<tag>]]` for each tag.
+5. Append category tags: `[[Category:Tech blog]]`; `[[Category:Blog:<year>]]` (from `date`, or the action's execution year if `date` is absent); `[[Category:<tag>]]` for each tag.
 6. Read the current wiki page content; if identical, skip the write (idempotency).
 7. If the page exists without `{{Auto-generated`, skip with a warning (manual page protection).
 8. Otherwise, write with an edit summary referencing the source commit.
@@ -80,7 +80,7 @@ For each `.md` with `wiki.publish: true`:
 - Pandoc-based markdown → wikitext rendering (h1 elements preserved)
 - Banner template injection
 - `Category:Tech blog` on every published page
-- `Category:Blog:<year>` from `date` field
+- `Category:Blog:<year>` from `date` field, falling back to the execution year
 - `Category:<tag>` per entry in `tags`
 - Idempotent writes
 - Manual-page protection

pipelines/tech/publish.py

@@ -41,9 +41,10 @@ def build_wikitext(fm: dict, body_wikitext: str, source_url: str, source_ref: st
     banner = f"{{{{Auto-generated|source={source_url}|commit={source_ref}}}}}"
     cat_parts = ["[[Category:Tech blog]]"]
     date_val = fm.get("date")
-    if date_val:
+    # Fall back to the action's execution year when no date is declared, so
-        year = str(date_val)[:4]
+    # every post still lands in a Blog:<year> category.
-        cat_parts.append(f"[[Category:Blog:{year}]]")
+    year = str(date_val)[:4] if date_val else str(datetime.now().year)
+    cat_parts.append(f"[[Category:Blog:{year}]]")
     for tag in (fm.get("tags") or []):
         cat_parts.append(f"[[Category:{tag}]]")
     categories = "\n".join(cat_parts) + "\n"