Merge pull request #4164 from gratipay/marky-markdown-package

Modify first h1/p just like npm does

.travis.yml

@@ -11,7 +11,7 @@ before_install:
  # Sometimes ya just halfta ...
  - git clone https://github.com/lloyd/yajl.git && cd yajl && git checkout 2.1.0 && ./configure && sudo make install && cd ..
 
- - npm install -g marky-markdown
+ - npm install marky-markdown
 cache:
   directories:
     - env/bin
@@ -23,6 +23,7 @@ install:
  - env/bin/pip install --upgrade ijson==2.3.0
 before_script:
  - echo "DATABASE_URL=dbname=gratipay" | tee -a tests/local.env local.env
+ - echo "REQUIRE_YAJL=true" | tee -a tests/local.env local.env
  - psql -U postgres -c 'CREATE DATABASE "gratipay";'
  - if [ "${TRAVIS_BRANCH}" = "master" -a "${TRAVIS_PULL_REQUEST}" = "false" ]; then rm -rfv tests/py/fixtures; fi
 script: LD_LIBRARY_PATH=/usr/local/lib make bgrun test doc

Makefile

@@ -63,7 +63,7 @@ pyflakes: env
 	$(env_bin)/pyflakes *.py bin gratipay tests
 
 test: test-schema
-	$(py_test) --cov gratipay ./tests/
+	$(py_test) -vv --cov gratipay ./tests/
 	@$(MAKE) --no-print-directory pyflakes
 
 pytest: env

bin/our-marky-markdown.js

@@ -0,0 +1,21 @@
+#!/usr/bin/env node
+// Extend marky-markdown.js to support the package argument:
+//  https://www.npmjs.com/package/marky-markdown#npm-packages
+
+var fs = require('fs')
+var path = require('path')
+var marky = require('marky-markdown')
+
+if (process.argv.length < 3) {
+  console.log('Usage:\n\nour-marky-markdown some.md pkg > some.html')
+  process.exit()
+}
+
+var filePath = path.resolve(process.cwd(), process.argv[2])
+
+fs.readFile(filePath, function (err, data) {
+  if (err) throw err
+  var package =  process.argv[3] ? JSON.parse(process.argv[3]) : null;
+  var html = marky(data.toString(), {package: package})
+  process.stdout.write(html)
+})

defaults.env

@@ -87,6 +87,8 @@ ASPEN_PROJECT_ROOT=.
 ASPEN_SHOW_TRACEBACKS=yes
 ASPEN_WWW_ROOT=www/
 
+# This is used in our Procfile. (PORT is also used but is provided by
+# Heroku; we don't set it ourselves in our app config.)
 # https://github.com/benoitc/gunicorn/issues/186
 GUNICORN_OPTS="--workers=1 --timeout=99999999"
 
@@ -96,5 +98,10 @@ TEAM_REVIEW_REPO=gratipay/test-gremlin
 TEAM_REVIEW_USERNAME=
 TEAM_REVIEW_TOKEN=
 
+# anything Postgres can interpret as an interval
 RESEND_VERIFICATION_THRESHOLD="3 minutes"
+
 RAISE_SIGNIN_NOTIFICATIONS=no
+
+# speeds up npm syncing; should be true on production and Travis
+REQUIRE_YAJL=false

gratipay/package_managers/readmes.py

@@ -24,8 +24,8 @@ def http_fetch(package_name):
     return r.json()
 
 
-def Syncer(db):
-    def sync(dirty, fetch=http_fetch):
+def Fetcher(db):
+    def fetch(dirty, fetch=http_fetch):
         """Update all info for one package.
         """
         log(dirty.name)
@@ -43,23 +43,58 @@ def sync(dirty, fetch=http_fetch):
         db.run('''
 
             UPDATE packages
-               SET readme=%s
+               SET readme_needs_to_be_processed=true
                  , readme_raw=%s
                  , readme_type=%s
              WHERE package_manager=%s
                AND name=%s
 
-        ''', ( markdown.marky(full['readme'])
-             , full['readme']
-             , 'x-markdown/npm'
+        ''', ( full['readme']
+             , 'x-markdown/marky'
+             , dirty.package_manager
+             , dirty.name
+              ))
+
+    return fetch
+
+
+def Processor(db):
+    def process(dirty):
+        """Processes the readme for a single page.
+        """
+        log(dirty.name)
+        raw = db.one( 'SELECT readme_raw FROM packages '
+                      'WHERE package_manager=%s and name=%s and readme_needs_to_be_processed'
+                    , (dirty.package_manager, dirty.name)
+                     )
+        if raw is None:
+            return
+        processed = markdown.render_like_npm(raw)
+        db.run('''
+
+            UPDATE packages
+               SET readme=%s
+                 , readme_needs_to_be_processed=false
+             WHERE package_manager=%s
+               AND name=%s
+
+        ''', ( processed
              , dirty.package_manager
              , dirty.name
               ))
 
-    return sync
+    return process
+
+
+def fetch(db):
+    dirty = db.all('SELECT package_manager, name '
+                   'FROM packages WHERE readme_raw IS NULL '
+                   'ORDER BY package_manager DESC, name DESC')
+    threaded_map(Fetcher(db), dirty, 4)
 
 
-def sync_all(db):
-    dirty = db.all('SELECT package_manager, name FROM packages WHERE readme_raw IS NULL '
+def process(db):
+    dirty = db.all('SELECT id, package_manager, name, description, readme_raw '
+                   'FROM packages WHERE readme_needs_to_be_processed'
                    'ORDER BY package_manager DESC, name DESC')
-    threaded_map(Syncer(db), dirty, 4)
+    threaded_map(Processor(db), dirty, 4)

gratipay/package_managers/sync.py

@@ -8,15 +8,21 @@
 import time
 import uuid
 
+from gratipay import wireup
 from gratipay.package_managers import readmes as _readmes
 
 
 log = lambda *a: print(*a, file=sys.stderr)
 NULL = uuid.uuid4().hex
 
 
-def import_ijson():
-    import ijson.backends.yajl2_cffi as ijson
+# helpers
+
+def import_ijson(env):
+    if env.require_yajl:
+        import ijson.backends.yajl2_cffi as ijson
+    else:
+        import ijson
     return ijson
 
 
@@ -50,10 +56,12 @@ def serialize_one(out, package):
     return 1
 
 
-def serialize(args):
+# cli subcommands
+
+def serialize(env, args, _):
     """Consume raw JSON from the npm registry and spit out CSV for Postgres.
     """
-    ijson = import_ijson()
+    ijson = import_ijson(env)
 
     path = args.path
     parser = ijson.parse(open(path))
@@ -97,9 +105,9 @@ def log_stats():
     log_stats()
 
 
-def upsert(args):
-    from gratipay import wireup
-    db = wireup.db(wireup.env())
+def upsert(env, args, db):
+    """Take a CSV file from stdin and load it into Postgres.
+    """
     fp = open(args.path)
     with db.get_cursor() as cursor:
         assert cursor.connection.encoding == 'UTF8'
@@ -128,19 +136,25 @@ def upsert(args):
         """)
 
 
-def readmes(args):
-    from gratipay import wireup
-    db = wireup.db(wireup.env())
-    _readmes.sync_all(db)
+def fetch_readmes(env, args, db):
+    _readmes.fetch(db)
+
+
+def process_readmes(env, args, db):
+    _readmes.process(db)
+
 
+# cli plumbing
 
 def parse_args(argv):
     p = argparse.ArgumentParser()
-    p.add_argument('command', choices=['serialize', 'upsert', 'readmes'])
+    p.add_argument('command', choices=['serialize', 'upsert', 'fetch-readmes', 'process-readmes'])
     p.add_argument('path', help='the path to the input file', nargs='?', default='/dev/stdin')
     return p.parse_args(argv)
 
 
 def main(argv=sys.argv):
+    env = wireup.env()
     args = parse_args(argv[1:])
-    globals()[args.command](args)
+    db = wireup.db(env)
+    globals()[args.command.replace('-', '_')](env, args, db)

gratipay/testing/__init__.py

@@ -3,8 +3,13 @@
 from __future__ import absolute_import, division, print_function, unicode_literals
 
 from decimal import Decimal
-from gratipay.models.participant import Participant
-from gratipay.models.team import Team
+
+import pytest
+from aspen import log_dammit
+
+from ..models.participant import Participant
+from ..models.team import Team
+from ..utils import markdown
 
 D = Decimal                     #:
 P = Participant.from_username   #:
@@ -45,3 +50,17 @@ def debug_http():
     requests_log = logging.getLogger("requests.packages.urllib3")
     requests_log.setLevel(logging.DEBUG)
     requests_log.propagate = True
+
+
+# Provide a decorator to skip tests when marky-markdown is missing.
+
+try:
+    markdown.render_like_npm('test')
+except OSError as exc:
+    MISSING_MARKY_MARKDOWN = True
+    log_dammit('Will skip marky-markdown-related tests because:', exc.args[0])
+else:
+    MISSING_MARKY_MARKDOWN = False
+
+def skipif_missing_marky_markdown(func):
+    return pytest.mark.skipif(MISSING_MARKY_MARKDOWN, reason="missing marky-markdown")(func)

gratipay/utils/markdown.py

@@ -1,21 +1,46 @@
+from __future__ import absolute_import, division, print_function, unicode_literals
+
 from subprocess import Popen, PIPE
 
-from markupsafe import Markup
+import json
 import misaka as m  # http://misaka.61924.nl/
+from markupsafe import Markup
 
 
 def render(markdown):
+    """Process markdown approximately the same way that GitHub used to.
+
+    (Note that as of November, 2016 they are migrating to CommonMark, so we are
+    starting to drift.)
+
+    """
     return Markup(m.html(
         markdown,
         extensions=m.EXT_AUTOLINK | m.EXT_STRIKETHROUGH | m.EXT_NO_INTRA_EMPHASIS,
         render_flags=m.HTML_SKIP_HTML | m.HTML_TOC | m.HTML_SMARTYPANTS | m.HTML_SAFELINK
     ))
 
 
-def marky(markdown):
+def render_like_npm(markdown, package=None):
     """Process markdown the same way npm does.
+
+    Package should be a dict representing the package. If it includes `name`
+    and `description` then the first h1 and paragraph will have a
+    'package-{name,description}-redundant' class added to them if they're
+    similar enough. If it includes `repository.url` then links will be changed
+    somehow. For details consult the docs and code:
+
+    https://github.com/npm/marky-markdown
+
     """
     if type(markdown) is unicode:
         markdown = markdown.encode('utf8')
-    marky = Popen(("marky-markdown", "/dev/stdin"), stdin=PIPE, stdout=PIPE)
-    return Markup(marky.communicate(markdown)[0])
+    cmd = ("bin/our-marky-markdown.js", "/dev/stdin")
+    if package:
+        cmd += (json.dumps(package),)
+    marky = Popen(cmd, stdin=PIPE, stdout=PIPE, stderr=PIPE)
+    out, err = marky.communicate(markdown)
+    if marky.wait() > 0:
+        raise OSError(err)
+    return out
+

gratipay/wireup.py

@@ -432,10 +432,8 @@ def env():
         TEAM_REVIEW_USERNAME            = unicode,
         TEAM_REVIEW_TOKEN               = unicode,
         RAISE_SIGNIN_NOTIFICATIONS      = is_yesish,
-        RESEND_VERIFICATION_THRESHOLD   = unicode, # anything Postgres can interpret as an interval
-
-        # This is used in our Procfile. (PORT is also used but is provided by
-        # Heroku; we don't set it ourselves in our app config.)
+        RESEND_VERIFICATION_THRESHOLD   = unicode,
+        REQUIRE_YAJL                    = is_yesish,
         GUNICORN_OPTS                   = unicode,
     )
 

requirements.txt

@@ -63,4 +63,6 @@
   ./vendor/ipaddress-1.0.16.tar.gz
 ./vendor/cryptography-1.3.2.tar.gz
 
+./vendor/ijson-2.3.tar.gz
+
 -e .

sql/branch.sql

@@ -0,0 +1 @@
+ALTER TABLE packages ADD COLUMN readme_needs_to_be_processed bool NOT NULL DEFAULT true;