atlass 1.5.0 → 1.7.0

package/README.md

@@ -1,317 +1,132 @@
 # atlass
 
-CLI to copy Jira issues and Confluence pages to Markdown, and push edits back to
-Confluence.
-
-Fetches an issue or page from Atlassian Cloud, converts its rich content to
-Markdown, writes a `.md` file with YAML frontmatter, and downloads any
-attachments alongside it. Confluence pages can be edited locally and updated
-back on the server.
+CLI for Atlassian Cloud. Views Jira issues in the terminal, copies Jira issues
+and Confluence pages to Markdown, pushes edits back, searches both, and shows
+Bitbucket pipeline runs.
 
 ## Install
 
 ```bash
-pnpm install
-pnpm build
-pnpm link --global   # exposes the `atlass` binary
+npm install -g atlass
 ```
 
-Or run from source without linking:
-
-```bash
-node dist/cli.mjs <command>
-```
+Or from source: `pnpm install && pnpm build && pnpm link --global`.
 
 ## Authentication
 
-Atlassian Cloud only. Auth uses your account email plus an API token
-(Basic auth). Create a token at
-`https://id.atlassian.com/manage-profile/security/api-tokens`.
+Create an API token at
+`https://id.atlassian.com/manage-profile/security/api-tokens`, then:
 
 ```bash
-atlass auth login
+atlass auth login    # prompts for site, email, and token
+atlass auth status
+atlass auth logout
 ```
 
-You are prompted for:
+Site and email go in `~/.config/atlass/config.json`. The token goes in the OS
+keyring. One account at a time.
 
-- site, e.g. `acme.atlassian.net`
-- account email
-- API token
-
-The login is verified against `/rest/api/3/myself` before anything is saved.
-The site and email are stored in `~/.config/atlass/config.json`
-(or `$XDG_CONFIG_HOME/atlass/config.json`). The API token is stored in the OS
-keyring (service `atlass`), never on disk.
+Bitbucket needs its own Bitbucket-scoped token with pipeline and workspace read
+scopes, since Atlassian tokens are scoped per product:
 
 ```bash
-atlass auth status   # show current site/email and whether a token is stored
-atlass auth logout   # remove config and delete the token from the keyring
+atlass bitbucket login    # prompts for workspace, default repo, and token
+atlass bitbucket status
+atlass bitbucket logout
 ```
 
-Only one account is supported at a time.
-
-## Usage
-
-### Copy a Jira issue
-
-```bash
-atlass jira copy PROJ-123
-atlass jira copy https://acme.atlassian.net/browse/PROJ-123
-atlass jira copy            # prompts for the key or URL
-```
-
-Accepts an issue key or any URL containing one. Writes `PROJ-123.md` to the
-current directory.
-
-### Update a Jira issue
-
-Copy an issue, edit the Markdown, then push the description back:
+## View
 
 ```bash
-atlass jira update PROJ-123.md
-atlass jira update                     # prompts for the file path
-atlass jira update file.md --dry-run   # show what would change, write nothing
-atlass jira update file.md --summary   # also push the H1 as the issue summary
+atlass jira view PROJ-123
+atlass jira view https://acme.atlassian.net/browse/PROJ-123
 ```
 
-The issue key comes from the file's frontmatter. The body is everything between
-the H1 and the `## Comments` section; the frontmatter, the H1, and the
-`## Comments` / `## Attachments` sections are not sent.
-
-Only the description is updated by default. Pass `--summary` to also push the H1
-as the new issue summary.
-
-Notes and safety:
-
-- The body is converted from Markdown to ADF. Only the standard constructs the
-  copy produces round-trip (headings, lists, task lists, code, blockquotes,
-  tables, rules, inline marks, links). Jira-specific content (panels, macros)
-  was flattened to plain Markdown on copy and cannot be rebuilt. When the live
-  description still contains such content, the update warns and asks for
-  confirmation before overwriting.
-- Jira has no page-style version number, so staleness is checked against the
-  frontmatter `updated` timestamp. If the issue changed since you copied it, the
-  update aborts so you can re-copy. `--force` overrides this and the data-loss
-  confirmation.
-- Image changes are not supported yet. External image URLs are kept as external
-  media, but a local image reference aborts the update (edit text only), and a
-  server-side image in the description is reported before it would be removed.
+Prints the issue to the terminal: fields, description, the last five comments
+(`--all-comments` for all of them), and attachment names. The status is colored
+by category and the Markdown is syntax highlighted, code blocks included; color
+turns off automatically when piping, or with `NO_COLOR`, leaving plain Markdown.
 
-### Copy a Confluence page
+## Copy
 
 ```bash
+atlass jira copy PROJ-123
+atlass jira copy https://acme.atlassian.net/browse/PROJ-123
 atlass confluence copy 123456
 atlass confluence copy https://acme.atlassian.net/wiki/spaces/DEV/pages/123456/Title
-atlass confluence copy     # prompts for the id or URL
 ```
 
-Accepts a numeric page id or a page URL. Writes `123456-title-slug.md` to the
-current directory.
+Writes `PROJ-123.md` or `123456-title-slug.md` to the current directory, with
+attachments in a sibling `<name>.assets/` folder. Pass `--out` with a directory
+or file path to write elsewhere. Existing files are overwritten.
 
-### Update a Confluence page
+Each file has YAML frontmatter, an H1, the body as Markdown, a `## Comments`
+section, and an `## Attachments` section. Inline images link to the local copy
+when they can be matched to an attachment.
 
-Copy a page, edit the Markdown, then push it back:
+## Update
+
+Edit a copied file, then push it back:
 
 ```bash
+atlass jira update PROJ-123.md
+atlass jira update PROJ-123.md --summary       # also push the H1 as the summary
 atlass confluence update 123456-title-slug.md
-atlass confluence update            # prompts for the file path
-atlass confluence update file.md --dry-run   # show what would change, write nothing
-atlass confluence update file.md --title     # also rename the page to the H1
+atlass confluence update file.md --title      # also push the H1 as the title
 atlass confluence update file.md -m "fix typo"
 ```
 
-The page id and version come from the file's frontmatter, so the file is
-self-describing. The body is everything between the H1 and the `## Comments`
-section; the frontmatter, the H1, and the `## Comments` / `## Attachments`
-sections are not sent as page content.
-
-Only the body is updated by default. Pass `--title` to also push the H1 as the
-new page title.
-
-Notes and safety:
-
-- The body is converted from Markdown to ADF. Only the standard constructs the
-  copy produces round-trip (headings, lists, task lists, code, blockquotes,
-  tables, rules, inline marks, links, images). Confluence-specific content
-  (panels, expands, macros, layouts) was flattened to plain Markdown on copy
-  and cannot be rebuilt. When the live page still contains such content, the
-  update warns and asks for confirmation before overwriting.
-- Before writing, the current server version is checked against the frontmatter
-  version. If the page changed since you copied it, the update aborts so you can
-  re-copy. `--force` overrides this and the data-loss confirmation.
-- Images referenced in the body are uploaded as attachments (matched by name and
-  size, so unchanged images are not re-uploaded). Local paths resolve relative
-  to the Markdown file; a missing local image aborts the update. External image
-  URLs are kept as external media.
-- Each update adds a version with the message `Updated via atlass` (override
-  with `--message`).
-
-### Search Jira issues
+Both accept `--dry-run` to preview and `--force` to skip the checks below. The
+body sent is everything between the H1 and `## Comments`.
 
-```bash
-atlass jira search "safari login"                 # free text
-atlass jira search --project PROJ --assignee me   # my open issues in PROJ
-atlass jira search --status "In Progress"
-atlass jira search --jql "project = PROJ AND labels = regression"
-atlass jira search                                # recent issues
-```
-
-Friendly filters (`--project`, `--assignee`, `--status`, text) are AND'd
-together and ordered by most recently updated. `--assignee me` maps to the
-current user. `--jql` takes a raw query and cannot be combined with the friendly
-filters. Prints one issue per line (`KEY  status  summary`); use `--json` for
-machine output. Only the first `--limit` results are shown (default 25, max
-100).
+- The update aborts if the issue or page changed on the server since the copy.
+- Panels, macros, expands, and layouts were flattened on copy and cannot be
+  rebuilt. If the live content still has them, the update asks before
+  overwriting.
+- Confluence uploads local images referenced in the body as attachments. Jira
+  update does not support image changes yet.
 
-### List Jira projects
+## Search
 
 ```bash
-atlass jira projects            # every project you can browse
-atlass jira projects pay        # filter by key or name
-atlass jira projects --json     # machine output
-```
-
-A discovery aid for the `--project` filter above: it fetches every project
-(paginated, ordered by key) and prints one per line as an aligned `KEY  Name`
-list. An optional query filters by key or name server-side. `--json` emits
-`{ key, name, id, type, url }` per project.
-
-### List Jira statuses
-
-```bash
-atlass jira statuses                 # every status on the site
-atlass jira statuses progress        # filter by name
-atlass jira statuses --project PROJ  # statuses used by one project
-atlass jira statuses --json          # machine output
-```
-
-A discovery aid for the `--status` filter above. Without `--project` it lists
-every status on the site; with `--project` it lists the statuses that project's
-issue types use (flattened to one list). An optional query filters by name
-(case-insensitive substring). Statuses are collapsed by name and category, then
-ordered by workflow lifecycle (To Do, then In Progress, then Done) and name.
-Prints an aligned `Name  Category` list; `--json` emits
-`{ name, id, category, categoryKey }` per status.
-
-### Search Confluence pages
-
-```bash
-atlass confluence search "onboarding"
-atlass confluence search --space DOCS
+atlass jira search "safari login"
+atlass jira search --project PROJ --assignee me --status "In Progress"
+atlass jira search --jql "project = PROJ AND labels = regression"
+atlass confluence search "onboarding" --space DOCS
 atlass confluence search --cql "label = runbook ORDER BY created DESC"
-atlass confluence search                          # recent pages
 ```
 
-Friendly mode always constrains to pages, so every result is copy-able. `--cql`
-takes a raw query and cannot be combined with `--space` or text. Prints one page
-per line (`id  space  title`); `--json` and `--limit` work as for Jira.
+Filters combine with AND and sort by most recently updated. `--jql` and `--cql`
+replace the friendly filters. `--limit` defaults to 25, max 100. `--json` prints
+machine output. Add `--copy` to pick results interactively and copy each one,
+with `--out` as the target directory.
 
-### Copy from search results
-
-Add `--copy` to any search to pick results interactively and copy each to
-Markdown (multi-select, needs an interactive terminal). With `--copy`, `--out`
-is a directory that every selected file is written into:
+Discovery aids for the filters:
 
 ```bash
-atlass jira search --project PROJ --copy --out ./tickets/
-atlass confluence search --space DOCS --copy
+atlass jira projects [query]
+atlass jira statuses [query] [--project PROJ]
 ```
 
-Copying continues on failure and reports a summary at the end.
-
-### Output location
-
-By default files are written to the current directory, named after the issue
-key (Jira) or `<id>-<title-slug>` (Confluence). Use `--out` to override:
+## Bitbucket pipelines
 
 ```bash
-atlass jira copy PROJ-123 --out ./tickets/          # directory: ./tickets/PROJ-123.md
-atlass jira copy PROJ-123 --out ./bug.md            # explicit file path
+atlass bitbucket pipelines                    # recent runs for the default repo
+atlass bitbucket pipelines --repo acme/web
+atlass bitbucket pipeline 124                 # one run and its steps
 ```
 
-Existing files are overwritten.
-
-## Output format
-
-Each file is:
-
-- YAML frontmatter with metadata
-- an H1 heading (issue summary / page title)
-- the body, converted from ADF to Markdown
-- a `## Comments` section (Jira comments / Confluence footer comments)
-- an `## Attachments` section linking every downloaded file
-
-Jira frontmatter: `key`, `type`, `status`, `assignee`, `reporter`, `priority`,
-`labels`, `created`, `updated`, `url`.
-
-Confluence frontmatter: `title`, `id`, `space`, `version`, `author`, `created`,
-`updated`, `url`.
-
-Example (Jira):
-
-```markdown
----
-key: "PROJ-123"
-type: "Bug"
-status: "In Progress"
-assignee: "Dana Scully"
-reporter: "Fox Mulder"
-priority: "High"
-labels:
-    - "regression"
-url: "https://acme.atlassian.net/browse/PROJ-123"
----
-
-# Login button does nothing on Safari
-
-Steps to reproduce...
-
-## Comments
-
-### Fox Mulder - 2025-07-01 10:30
-
-Reproduced on 17.5.
-
-## Attachments
-
-- [screenshot.png](PROJ-123.assets/screenshot.png)
-```
-
-### Attachments
-
-All attachments are downloaded into a sibling `<name>.assets/` folder and
-listed under `## Attachments`. Inline images are linked to the local copy where
-they can be matched:
-
-- Confluence media nodes carry the attachment file id, so inline images resolve
-  reliably.
-- Jira media nodes usually expose only a filename, so inline images resolve when
-  that filename matches an attachment; otherwise they render as
-  `[embedded media: ...]` and the file is still captured in the attachments
-  folder and section.
+`--repo` takes `workspace/slug` or a bare slug under the configured workspace.
+`--limit` and `--json` work as for search.
 
 ## Development
 
 ```bash
-pnpm test      # run unit tests
-pnpm check     # format, lint, typecheck (add --fix to auto-fix)
+pnpm test      # unit tests
+pnpm check     # format, lint, typecheck (--fix to auto-fix)
 pnpm build     # build dist/cli.mjs
 pnpm dev       # build in watch mode
 ```
 
-### Layout
-
-- `src/cli.ts` command wiring (commander)
-- `src/commands/` `auth`, `jira`, `confluence` command handlers
-- `src/api/` fetch client, Jira and Confluence endpoints, attachment up/download
-- `src/adf/` ADF to Markdown and Markdown to ADF converters (unit tested)
-- `src/markdown/` frontmatter, comments, attachments, media resolver, update source
-- `src/config.ts`, `src/credentials.ts` config file and keyring
-- `src/util/` key/id parsing and output path resolution
-
-The ADF to Markdown converter in `src/adf/to-markdown.ts` is a single hand
-rolled walker shared by both commands. Confluence page bodies are requested as
-`atlas_doc_format` so they flow through the same converter as Jira. The reverse
-direction, `src/adf/from-markdown.ts`, tokenizes Markdown with `marked` and
-emits ADF for the Confluence update command; it covers the same clean subset the
-copy produces.
+Commands live in `src/commands/`, API clients in `src/api/`, and the ADF
+converters in `src/adf/`.