flowcvcli 0.5.2__tar.gz → 0.7.0__tar.gz

{flowcvcli-0.5.2/flowcvcli.egg-info → flowcvcli-0.7.0}/PKG-INFO

@@ -1,7 +1,7 @@
 Metadata-Version: 2.4
 Name: flowcvcli
-Version: 0.5.2
-Summary: Control a FlowCV resume from the command line or Python — content, design, templates, photo, publish and PDF export — via FlowCV's private JSON API. Zero dependencies.
+Version: 0.7.0
+Summary: Control a FlowCV resume from the command line or Python — content, design, templates, photo, publish and PDF export — via FlowCV's private JSON API.
 Author: dannyota
 License-Expression: MIT
 Project-URL: Homepage, https://github.com/dannyota/flowcvcli
@@ -26,6 +26,7 @@ Classifier: Topic :: Utilities
 Requires-Python: >=3.8
 Description-Content-Type: text/markdown
 License-File: LICENSE
+Requires-Dist: curl_cffi>=0.7
 Dynamic: license-file
 
 # flowcvcli
@@ -36,10 +37,14 @@ Dynamic: license-file
 
 Control a [FlowCV](https://flowcv.com) resume from the **command line** or from
 **Python** — content, header & links, **customization**, **templates**,
-**avatar**, reorder/hide, multi-resume management, **backup/restore**, publish, and **PDF export**.
+**avatar**, reorder/hide, multi-resume management, **backup/restore**, publish,
+**PDF export**, **resume-as-code** (`pull`/`push` a git-diffable Markdown tree),
+**JSON Resume** import/export, `$EDITOR` editing, auto-snapshots before
+destructive ops, and machine-readable **`--json`** output for every command.
 It drives FlowCV's private JSON API (the same calls the web app makes), so it
-works for any FlowCV resume with your own session. **Zero dependencies** (Python
-standard library only), so it's easy to drop into scripts and LLM agents.
+works for any FlowCV resume with your own session. One tiny dependency
+(`curl_cffi`, only for the login TLS fingerprint); everything else is the Python
+standard library, so it drops easily into scripts and LLM agents.
 
 > Unofficial and not affiliated with FlowCV. It uses FlowCV's undocumented
 > internal API and may break if that changes; use it with your own account and at
@@ -102,7 +107,8 @@ flowcv add work --set title="Engineer" --set company="Acme" \
        --set start=01/2022 --set end=Present --text $'- Did a measurable thing.'
 flowcv add custom2 --section-name "Open Source" --icon code --text "…"   # heading+icon at creation
 flowcv desc work <id> --file role.md           # rich text; --field is optional (auto: profile=text, skill=infoHtml)
-flowcv date publication <id> --year 2018       # structured date (year-only by default)
+flowcv edit work <id>                          # open the entry's rich text as markdown in $EDITOR, save on change
+flowcv date publication <id> --year 2018       # structured date; merges (only passed parts change), --clear resets
 flowcv field work <id> employer --text "Acme Corp"
 flowcv rm work <id>
 
@@ -113,6 +119,7 @@ flowcv rename-section skill "Core Skills"
 flowcv section-icon skill head-side-brain
 flowcv rm-section custom1 --yes           # delete a section + its entries
 flowcv reorder-sections profile work skill education   # set one-column order (no args = print current)
+flowcv reorder-sections work skill --layout two --side left   # two-column: order one column at a time
 
 # header details & links (links are social entries: orcid, googlescholar, github…)
 flowcv pd jobTitle --text "Security Leader"
@@ -124,26 +131,47 @@ flowcv avatar set https://example.com/me.png   # upload from URL or file
 flowcv avatar on | off | remove
 
 # styling (a delta into resume.customization) and templates
-flowcv customize font.fontFamily "Source Sans Pro"
+flowcv customize                     # no value = print the current customization tree
+flowcv customize font                #   filter to a subtree (leaves as dot.path = value)
+flowcv customize font.fontFamily "Source Sans Pro"   # with a value = set that dot-path
 flowcv customize colors.basic.single '"#0e374e"'
+flowcv icons                         # list commonly-used section iconKeys
 flowcv templates                     # lists each as [free] / [PAID] (paid needs a subscription)
 flowcv apply-template <templateId>   # warns first if the template is paid
 
 # render & share
-flowcv download -o resume.pdf        # the rendered PDF
+flowcv download -o resume.pdf        # the rendered PDF (--pages N caps rendered pages; default 10)
 flowcv download --token <webToken> -o out.pdf   # any PUBLIC resume by its share token (no auth)
 flowcv share | publish | unpublish
 
 # backup / restore
 flowcv export -o backup.json          # full resume snapshot (JSON) — keep one before big edits
 flowcv import backup.json             # restore the snapshot into a NEW resume (non-destructive)
+flowcv export --format jsonresume -o me.json   # export to the jsonresume.org schema
+flowcv import --format jsonresume me.json      # build a NEW resume from a JSON Resume doc
+flowcv backups                        # list auto-snapshots (rm-section / delete-resume snapshot first; --no-backup opts out)
 
 flowcv login                          # refresh the cached session
+flowcv doctor                         # diagnose auth/session/setup (--offline skips the live probe)
 flowcv md2html --file role.md         # preview HTML (offline)
 ```
 
-Any command takes `--resume-id <id>` to target a specific resume. (From source,
-replace `flowcv` with `python3 flowcv.py`.)
+Any command takes `--resume-id <id>` to target a specific resume, and `--json`
+(before or after the subcommand) to emit exactly one machine-readable JSON
+document instead of human text — errors become `{"error", "type"}` on stdout
+with exit 1. (From source, replace `flowcv` with `python3 flowcv.py`.)
+
+### Resume as code
+
+```bash
+flowcv pull [dir]            # materialize the resume as a Markdown tree (default ./resume)
+# edit the files: personal.md, NN-<section>/_section.md, NN-<id8>.md (frontmatter + body)
+flowcv push [dir] --dry-run  # review the diff (last-writer-wins; no 3-way merge)
+flowcv push [dir]            # apply ONLY changes (edits, adds, deletes, reorder, renames)
+```
+
+`pull` then an immediate `push` is a no-op. Full layout, frontmatter rules and
+round-trip caveats: [`docs/PULLPUSH.md`](docs/PULLPUSH.md).
 
 ## Library (for scripts & LLM agents)
 
@@ -166,12 +194,29 @@ fc.rename_section("skill", "Core Skills"); fc.delete_section("custom1")
 fc.hide_entry("work", "id", hidden=True)
 new_id = fc.create_resume("Second Resume")          # or fc.duplicate_resume()
 fc.rename_resume("New Title"); fc.delete_resume()    # delete is permanent
-fc.set_date("publication", "id", year=2018)         # structured date (year-only)
+fc.set_date("publication", "id", year=2018)         # structured date; merges (only passed parts change)
 
 # backup / restore
 import json
 json.dump(fc.export_resume(), open("backup.json", "w"))   # full snapshot
 new_id = fc.import_resume(json.load(open("backup.json")))  # restore into a NEW resume
+
+# JSON Resume interop (jsonresume.org schema)
+from flowcvcli import to_jsonresume, from_jsonresume
+jr = to_jsonresume(fc.export_resume())                     # FlowCV -> JSON Resume dict
+fc.import_resume(from_jsonresume(jr, base=fc.get_resume())) # JSON Resume -> a NEW resume
+
+# many edits? batch caches the read (1 GET instead of N against the rate limit)
+with fc.batch():
+    fc.set_field("work", "id", "employer", "Acme Corp")
+    fc.set_date("publication", "id", year=2018)
+
+# errors are normal exceptions (they never SystemExit your process):
+from flowcvcli import FlowCVError, AuthError, RateLimitError, NotFoundError
+try:
+    fc.find_entry(fc.get_resume(), "work", "bad-id")
+except NotFoundError as e:
+    print(e)                                    # all subclass FlowCVError
 ```
 
 ### Build → render → check → improve
@@ -192,6 +237,10 @@ feedback loop for building a resume from raw info.
 | `**bold**` / `***bold-italic***` inline | `<strong>` / `<strong><em>…</em></strong>` |
 | `[text](url)` inline | link |
 
+Bold markers inside an already-bold line or `**span**` are redundant and
+collapse to plain text — this keeps markdown ⇄ HTML round-trips stable
+(`flowcv edit`, `pull`/`push`).
+
 ## How it works
 
 - **Read-modify-write**: edits fetch the resume, change one part, and send it
@@ -209,16 +258,21 @@ feedback loop for building a resume from raw info.
 ```
 flowcvcli/             # the package (import flowcvcli)
   config.py            #   resolve resume id + auth from .env / env vars
-  client.py            #   HTTP, login, cookie-jar session, retry, get_resume
-  markup.py            #   markdown <-> FlowCV rich-text HTML
+  client.py            #   HTTP, login, cookie-jar session, retry, batch cache
+  errors.py            #   FlowCVError hierarchy (Auth/RateLimit/NotFound/Api)
+  markup.py            #   markdown <-> FlowCV rich-text HTML (both directions)
   content.py           #   sections & entries (add/edit/reorder/hide/sections)
   personal.py          #   header details & links
   customization.py     #   styling deltas & templates
+  jsonresume.py        #   JSON Resume (jsonresume.org) import/export
+  repo.py              #   resume-as-code: pull/push Markdown tree
   photo.py             #   avatar upload / toggle
-  resume.py            #   list, create/duplicate/rename/delete, download, publish
+  resume.py            #   list, create/duplicate/rename/delete, download, publish, snapshots
   api.py               #   FlowCV = Client + all mixins
   cli.py / __main__.py #   the `flowcv` command
+tests/                 # offline unit tests: python3 -m unittest discover -s tests -t .
 docs/API.md            # reverse-engineered API reference
+docs/PULLPUSH.md       # resume-as-code design (layout, frontmatter, diff rules)
 docs/RENDERING.md      # how the editor renders the preview & debounces saves
 flowcv.py              # source-tree entry point (python3 flowcv.py …)
 ```

{flowcvcli-0.5.2 → flowcvcli-0.7.0}/README.md

@@ -6,10 +6,14 @@
 
 Control a [FlowCV](https://flowcv.com) resume from the **command line** or from
 **Python** — content, header & links, **customization**, **templates**,
-**avatar**, reorder/hide, multi-resume management, **backup/restore**, publish, and **PDF export**.
+**avatar**, reorder/hide, multi-resume management, **backup/restore**, publish,
+**PDF export**, **resume-as-code** (`pull`/`push` a git-diffable Markdown tree),
+**JSON Resume** import/export, `$EDITOR` editing, auto-snapshots before
+destructive ops, and machine-readable **`--json`** output for every command.
 It drives FlowCV's private JSON API (the same calls the web app makes), so it
-works for any FlowCV resume with your own session. **Zero dependencies** (Python
-standard library only), so it's easy to drop into scripts and LLM agents.
+works for any FlowCV resume with your own session. One tiny dependency
+(`curl_cffi`, only for the login TLS fingerprint); everything else is the Python
+standard library, so it drops easily into scripts and LLM agents.
 
 > Unofficial and not affiliated with FlowCV. It uses FlowCV's undocumented
 > internal API and may break if that changes; use it with your own account and at
@@ -72,7 +76,8 @@ flowcv add work --set title="Engineer" --set company="Acme" \
        --set start=01/2022 --set end=Present --text $'- Did a measurable thing.'
 flowcv add custom2 --section-name "Open Source" --icon code --text "…"   # heading+icon at creation
 flowcv desc work <id> --file role.md           # rich text; --field is optional (auto: profile=text, skill=infoHtml)
-flowcv date publication <id> --year 2018       # structured date (year-only by default)
+flowcv edit work <id>                          # open the entry's rich text as markdown in $EDITOR, save on change
+flowcv date publication <id> --year 2018       # structured date; merges (only passed parts change), --clear resets
 flowcv field work <id> employer --text "Acme Corp"
 flowcv rm work <id>
 
@@ -83,6 +88,7 @@ flowcv rename-section skill "Core Skills"
 flowcv section-icon skill head-side-brain
 flowcv rm-section custom1 --yes           # delete a section + its entries
 flowcv reorder-sections profile work skill education   # set one-column order (no args = print current)
+flowcv reorder-sections work skill --layout two --side left   # two-column: order one column at a time
 
 # header details & links (links are social entries: orcid, googlescholar, github…)
 flowcv pd jobTitle --text "Security Leader"
@@ -94,26 +100,47 @@ flowcv avatar set https://example.com/me.png   # upload from URL or file
 flowcv avatar on | off | remove
 
 # styling (a delta into resume.customization) and templates
-flowcv customize font.fontFamily "Source Sans Pro"
+flowcv customize                     # no value = print the current customization tree
+flowcv customize font                #   filter to a subtree (leaves as dot.path = value)
+flowcv customize font.fontFamily "Source Sans Pro"   # with a value = set that dot-path
 flowcv customize colors.basic.single '"#0e374e"'
+flowcv icons                         # list commonly-used section iconKeys
 flowcv templates                     # lists each as [free] / [PAID] (paid needs a subscription)
 flowcv apply-template <templateId>   # warns first if the template is paid
 
 # render & share
-flowcv download -o resume.pdf        # the rendered PDF
+flowcv download -o resume.pdf        # the rendered PDF (--pages N caps rendered pages; default 10)
 flowcv download --token <webToken> -o out.pdf   # any PUBLIC resume by its share token (no auth)
 flowcv share | publish | unpublish
 
 # backup / restore
 flowcv export -o backup.json          # full resume snapshot (JSON) — keep one before big edits
 flowcv import backup.json             # restore the snapshot into a NEW resume (non-destructive)
+flowcv export --format jsonresume -o me.json   # export to the jsonresume.org schema
+flowcv import --format jsonresume me.json      # build a NEW resume from a JSON Resume doc
+flowcv backups                        # list auto-snapshots (rm-section / delete-resume snapshot first; --no-backup opts out)
 
 flowcv login                          # refresh the cached session
+flowcv doctor                         # diagnose auth/session/setup (--offline skips the live probe)
 flowcv md2html --file role.md         # preview HTML (offline)
 ```
 
-Any command takes `--resume-id <id>` to target a specific resume. (From source,
-replace `flowcv` with `python3 flowcv.py`.)
+Any command takes `--resume-id <id>` to target a specific resume, and `--json`
+(before or after the subcommand) to emit exactly one machine-readable JSON
+document instead of human text — errors become `{"error", "type"}` on stdout
+with exit 1. (From source, replace `flowcv` with `python3 flowcv.py`.)
+
+### Resume as code
+
+```bash
+flowcv pull [dir]            # materialize the resume as a Markdown tree (default ./resume)
+# edit the files: personal.md, NN-<section>/_section.md, NN-<id8>.md (frontmatter + body)
+flowcv push [dir] --dry-run  # review the diff (last-writer-wins; no 3-way merge)
+flowcv push [dir]            # apply ONLY changes (edits, adds, deletes, reorder, renames)
+```
+
+`pull` then an immediate `push` is a no-op. Full layout, frontmatter rules and
+round-trip caveats: [`docs/PULLPUSH.md`](docs/PULLPUSH.md).
 
 ## Library (for scripts & LLM agents)
 
@@ -136,12 +163,29 @@ fc.rename_section("skill", "Core Skills"); fc.delete_section("custom1")
 fc.hide_entry("work", "id", hidden=True)
 new_id = fc.create_resume("Second Resume")          # or fc.duplicate_resume()
 fc.rename_resume("New Title"); fc.delete_resume()    # delete is permanent
-fc.set_date("publication", "id", year=2018)         # structured date (year-only)
+fc.set_date("publication", "id", year=2018)         # structured date; merges (only passed parts change)
 
 # backup / restore
 import json
 json.dump(fc.export_resume(), open("backup.json", "w"))   # full snapshot
 new_id = fc.import_resume(json.load(open("backup.json")))  # restore into a NEW resume
+
+# JSON Resume interop (jsonresume.org schema)
+from flowcvcli import to_jsonresume, from_jsonresume
+jr = to_jsonresume(fc.export_resume())                     # FlowCV -> JSON Resume dict
+fc.import_resume(from_jsonresume(jr, base=fc.get_resume())) # JSON Resume -> a NEW resume
+
+# many edits? batch caches the read (1 GET instead of N against the rate limit)
+with fc.batch():
+    fc.set_field("work", "id", "employer", "Acme Corp")
+    fc.set_date("publication", "id", year=2018)
+
+# errors are normal exceptions (they never SystemExit your process):
+from flowcvcli import FlowCVError, AuthError, RateLimitError, NotFoundError
+try:
+    fc.find_entry(fc.get_resume(), "work", "bad-id")
+except NotFoundError as e:
+    print(e)                                    # all subclass FlowCVError
 ```
 
 ### Build → render → check → improve
@@ -162,6 +206,10 @@ feedback loop for building a resume from raw info.
 | `**bold**` / `***bold-italic***` inline | `<strong>` / `<strong><em>…</em></strong>` |
 | `[text](url)` inline | link |
 
+Bold markers inside an already-bold line or `**span**` are redundant and
+collapse to plain text — this keeps markdown ⇄ HTML round-trips stable
+(`flowcv edit`, `pull`/`push`).
+
 ## How it works
 
 - **Read-modify-write**: edits fetch the resume, change one part, and send it
@@ -179,16 +227,21 @@ feedback loop for building a resume from raw info.
 ```
 flowcvcli/             # the package (import flowcvcli)
   config.py            #   resolve resume id + auth from .env / env vars
-  client.py            #   HTTP, login, cookie-jar session, retry, get_resume
-  markup.py            #   markdown <-> FlowCV rich-text HTML
+  client.py            #   HTTP, login, cookie-jar session, retry, batch cache
+  errors.py            #   FlowCVError hierarchy (Auth/RateLimit/NotFound/Api)
+  markup.py            #   markdown <-> FlowCV rich-text HTML (both directions)
   content.py           #   sections & entries (add/edit/reorder/hide/sections)
   personal.py          #   header details & links
   customization.py     #   styling deltas & templates
+  jsonresume.py        #   JSON Resume (jsonresume.org) import/export
+  repo.py              #   resume-as-code: pull/push Markdown tree
   photo.py             #   avatar upload / toggle
-  resume.py            #   list, create/duplicate/rename/delete, download, publish
+  resume.py            #   list, create/duplicate/rename/delete, download, publish, snapshots
   api.py               #   FlowCV = Client + all mixins
   cli.py / __main__.py #   the `flowcv` command
+tests/                 # offline unit tests: python3 -m unittest discover -s tests -t .
 docs/API.md            # reverse-engineered API reference
+docs/PULLPUSH.md       # resume-as-code design (layout, frontmatter, diff rules)
 docs/RENDERING.md      # how the editor renders the preview & debounces saves
 flowcv.py              # source-tree entry point (python3 flowcv.py …)
 ```

{flowcvcli-0.5.2 → flowcvcli-0.7.0}/docs/API.md

@@ -37,7 +37,7 @@ on the same cookie jar → the jar now holds the authenticated `flowcvsidapp`.
 |---|---|---|---|
 | GET | `/resumes/all` | — | `data.resumes[]` (id, title, webToken, webResumeLive, order, …) |
 | GET | `/resumes/{resumeId}` | — | `data.resume` (full resume object) |
-| POST | `/resumes/create` | `{clientResume: {…full resume object…}}` | create a resume. The body must be a **complete** resume object (every NOT-NULL column), so the reliable way is to **clone a full existing resume** (`GET /resumes/{id}`), reassign `id`+`uuid`, set `title`, empty `content` (or keep it for a duplicate), and **drop** `webToken`/`feedbackToken`/`createdAt`/`updatedAt` (server regenerates). A hand-built partial body fails with Postgres `23502` (not-null). Note: the one-resume free-plan cap is **not** enforced on this endpoint. |
+| POST | `/resumes/create` | `{clientResume: {…full resume object…}}` | create a resume. The body must be a **complete** resume object (every NOT-NULL column), so the reliable way is to **clone a full existing resume** (`GET /resumes/{id}`), reassign `id`+`uuid`, set `title`, empty `content` (or keep it for a duplicate), and **drop** `webToken`/`feedbackToken`/`createdAt`/`updatedAt` (server regenerates). A hand-built partial body fails with Postgres `23502` (not-null). Note: the one-resume free-plan cap is **not** enforced on this endpoint. **The server IGNORES the client-supplied `id` and mints its own** — the created resume comes back in `data.resume`; read the real id from there (using the client id yields `400 reloadClient:true` on every follow-up). `webResumeLive` IS honored from the clone: cloning a live resume publishes the copy under a fresh `webToken` unless you set it `false`. |
 | POST | `/resumes/duplicate` | `{resumeId}` | native duplicate — but returned a generic error in testing; duplicating via `create` (clone, keep `content`) is what this tool does instead. |
 | PATCH | `/resumes/rename_resume` | `{resumeId, resumeTitle}` | rename a resume |
 | DELETE | `/resumes/delete_resume?resumeId={id}` | — | **permanently delete** a resume (irreversible) |

flowcvcli-0.7.0/docs/PULLPUSH.md

@@ -0,0 +1,87 @@
+# Resume as code — `flowcv pull` / `flowcv push`
+
+Materialize a FlowCV resume as an editable, git-diffable tree of Markdown files,
+edit it locally, then push only the changes back. Last-writer-wins; there is no
+3-way merge — `push --dry-run` is the review tool.
+
+## File layout (`./resume/` by default)
+
+```
+resume/
+  .flowcv.json                 # manifest: resumeId, pulledAt, {entryId: relpath}
+  personal.md                  # personalDetails as frontmatter, NO body
+  00-profile/
+    _section.md                # frontmatter: displayName, iconKey, sectionType
+    00-<id8>.md                # one entry: frontmatter (scalars) + markdown body
+  01-work/
+    _section.md
+    00-<id8>.md
+    01-<id8>.md
+  …
+```
+
+- Section dirs are `NN-<sectionId>`, `NN` = 2-digit index in the resume's
+  `content` order (display order). Entry files are `NN-<first 8 of id>.md`,
+  `NN` = the entry's index within the section (its order).
+- **Filenames are cosmetic** — ordering + a human hint. The authoritative entry
+  id is the `id` in each file's frontmatter; the manifest maps every full id to
+  its file. `push` addresses entries by that id, never by the filename.
+
+## Frontmatter format (tiny, no PyYAML)
+
+Plain `key: value` lines between `---` fences. A value is written **verbatim**
+when it is a string that survives the round-trip; otherwise it is
+`json.dumps`-encoded (numbers, bools, null, lists, dicts, and any string that
+would otherwise parse as JSON, has leading/trailing space, or contains a
+newline). Parsing is the exact inverse: `json.loads` the value; on failure it is
+a plain string. So `startDateNew: 01/2022` and `endDateNew: Present` stay plain,
+`isHidden: false` is a bool, `endDateNew: "2019"` is forced-quoted to stay a
+string, and `date: {"year": "2018", …}` is JSON. `personal.md` puts the whole
+`personalDetails` object (scalars, `social`, `detailsOrder`, `photo`, …) into
+frontmatter this way.
+
+## Entry body
+
+The body is `html_to_md` of the entry's rich-text field (`text` for profile,
+`infoHtml` for skill, else `description`; see `content.rich_field`). All other
+fields are frontmatter. On push the body is compared in **markdown** space
+(`html_to_md(live)` vs the file), and only re-encoded with `md_to_html` when it
+differs — this is what makes the round-trip stable despite `md_to_html` /
+`html_to_md` not being byte-inverse on non-canonical server HTML.
+
+## `push` — diff one snapshot, apply only changes
+
+One `get_resume()` under `batch()`; every diff is computed against that snapshot,
+then applied with low-level writers (no re-reads):
+
+| Change detected | API call |
+|---|---|
+| entry frontmatter field and/or body differs | `save_entry` (merged onto the live entry, `updatedAt` bumped) |
+| entry file deleted (live id not referenced by any file) | `delete_entry` |
+| entry file with **no** `id` in frontmatter | create (two `save_entry` calls); new id written back into the file + manifest |
+| section entry order (file `NN` order) differs | `reorder_entries` |
+| `_section.md` displayName / iconKey differs | `rename_section` / `set_section_icon` |
+| `personal.md` differs | `save_personal` (merged onto live `personalDetails`) |
+
+Semantics: **overlay, never destroy.** Frontmatter/personal keys you delete from
+a file are left untouched on the server; `push` never deletes a section or the
+resume, and never creates a section (folders whose sectionId isn't live are
+skipped). A frontmatter `id` that matches no live entry is skipped with a note.
+`--dry-run` computes and prints/`--json`-emits the actions without applying any.
+
+## Round-trip guarantee
+
+`pull` then an immediate `push` yields **zero actions**. `pull` writes every live
+field verbatim (frontmatter serialization is an exact inverse) and the body as
+`html_to_md(live)`; `push` compares frontmatter parsed-value-equal and the body
+in markdown space, so nothing differs. Caveats that keep this true:
+
+- Bodies are diffed as markdown, not re-encoded HTML (so `md_to_html`/`html_to_md`
+  non-canonical drift never triggers a write). After a real body edit is pushed,
+  the server HTML is canonical and every later round-trip stays stable
+  (`md_to_html(html_to_md(md_to_html(x))) == md_to_html(x)`; see `markup.py`).
+- `push` overlays onto the live object, so server-only keys absent from a file
+  (e.g. a new `personalDetails` field) are preserved, not diffed.
+- The manifest `pulledAt` is **informational only** — there is no conflict
+  detection. If the server changed since your pull, `push` overwrites with your
+  local values (last-writer-wins); run `push --dry-run` first to review.

flowcvcli-0.7.0/flowcvcli/__init__.py

@@ -0,0 +1,13 @@
+"""flowcvcli — control a FlowCV resume from Python or the command line."""
+from .api import FlowCV
+from .config import Config
+from .content import SECTION_META, label_of
+from .errors import (ApiError, AuthError, FlowCVError, NotFoundError,
+                     RateLimitError)
+from .jsonresume import from_jsonresume, to_jsonresume
+from .markup import html_to_md, html_to_text, md_to_html
+
+__all__ = ["FlowCV", "Config", "SECTION_META", "label_of", "md_to_html", "html_to_text",
+           "html_to_md", "to_jsonresume", "from_jsonresume",
+           "FlowCVError", "AuthError", "RateLimitError", "NotFoundError", "ApiError"]
+__version__ = "0.7.0"