examples : make badge.py work without GH token

If the user doesn't supply GH token, try to fetch and parse profile
information from public HTML

examples/README.md

@@ -45,13 +45,15 @@ $ ./demo.py save
 # `badge.py`
 Creates a badge-like tag for a GitHub user. Some examples are [octocat](https://ggtag.io/?i=%5Cr10%2C25%2C110%2C110%5CI15%2C30%2C100%2C100%2C1%2Chttps%3A%2F%2Favatars.githubusercontent.com%2Fu%2F583231%3Fv%3D4%5Ct140%2C50%2C5%2CThe%20Octocat%5Ct140%2C80%2C2%2Cgithub.com%2Foctocat%5Ca13%2C156%2C16%2Cmap-marker-alt%5Ct33%2C158%2C2%2CSan%20Francisco%5Ca13%2C183%2C16%2Cbuilding%5Ct33%2C185%2C2%2C%40github%5Ca180%2C154%2C16%2Clink%5Ct202%2C158%2C2%2Chttps%3A%2F%2Fgithub.blog%5Ca180%2C185%2C16%2Cenvelope%5Ct202%2C185%2C2%2Coctocat%40github.com), [antirez](https://ggtag.io/?i=%5Cr10%2C25%2C110%2C110%5CI15%2C30%2C100%2C100%2C1%2Chttps%3A%2F%2Favatars.githubusercontent.com%2Fu%2F65632%3Fv%3D4%5Ct140%2C40%2C5%2CSalvatore%5Ct140%2C70%2C5%2CSanfilippo%5Ct140%2C110%2C2%2Cgithub.com%2Fantirez%5Ca13%2C156%2C16%2Cmap-marker-alt%5Ct33%2C158%2C2%2CCatania%2CSicily%2CItaly%5Ca13%2C183%2C16%2Cbuilding%5Ct33%2C185%2C2%2CRedis%20Labs%5Ca180%2C154%2C16%2Clink%5Ct202%2C158%2C2%2Chttp%3A%2F%2Finvece.org%5Ca180%2C185%2C16%2Cenvelope%5Ct202%2C185%2C2%2Cantirez%40gmail.com), [ggerganov](https://ggtag.io/?i=%5Cr10%2C25%2C110%2C110%5CI15%2C30%2C100%2C100%2C1%2Chttps%3A%2F%2Favatars.githubusercontent.com%2Fu%2F1991296%3Fv%3D4%5Ct140%2C40%2C5%2CGeorgi%5Ct140%2C70%2C5%2CGerganov%5Ct140%2C110%2C2%2Cgithub.com%2Fggerganov%5Ca13%2C156%2C16%2Cmap-marker-alt%5Ct33%2C158%2C2%2CSofia%2C%20Bulgaria%5Ca13%2C183%2C16%2Cbuilding%5Ct33%2C185%2C2%2C%40viewray-inc%20%5Ca180%2C154%2C16%2Clink%5Ct202%2C158%2C2%2Chttps%3A%2F%2Fggerganov.com%5Ca180%2C185%2C16%2Cenvelope%5Ct202%2C185%2C2%2Cggerganov%40gmail.com), [rgerganov](https://ggtag.io/?i=%5Cr10%2C25%2C110%2C110%5CI15%2C30%2C100%2C100%2C1%2Chttps%3A%2F%2Favatars.githubusercontent.com%2Fu%2F271616%3Fv%3D4%5Ct140%2C40%2C5%2CRadoslav%5Ct140%2C70%2C5%2CGerganov%5Ct140%2C110%2C2%2Cgithub.com%2Frgerganov%5Ca13%2C156%2C16%2Cmap-marker-alt%5Ct33%2C158%2C2%2CSofia%2C%20Bulgaria%5Ca13%2C183%2C16%2Cbuilding%5Ct33%2C185%2C2%2C%40vmware%5Ca180%2C154%2C16%2Clink%5Ct202%2C158%2C2%2Chttps%3A%2F%2Fxakcop.com%5Ca180%2C185%2C16%2Cenvelope%5Ct202%2C185%2C2%2Crgerganov%40gmail.com).
 
-You need to create a [personal access token](https://docs.github.com/en/authentication/keeping-your-account-and-data-secure/creating-a-personal-access-token) on GitHub and export it as `GITHUB_TOKEN` environment variable.
+For best results, we suggest to create a [personal access token](https://docs.github.com/en/authentication/keeping-your-account-and-data-secure/creating-a-personal-access-token) on GitHub and export it as `GITHUB_TOKEN` environment variable.
 ```
 $ export GITHUB_TOKEN=<your-github-token>
 $ ./badge.py <github-username>
 # You can replace the profile picture with QR code, so you can program the tag with sound:
 $ ./badge.py --qrcode <github-username>
 ```
+If you don't set `GITHUB_TOKEN`, the script will try to fetch and parse the profile information from GitHub website, which is less reliable.
+
 # `img.py`
 Creates a tag with PNG image specified either as local file or URL:
 ```

examples/badge.py

@@ -5,8 +5,67 @@
 import ggtag
 import urllib.request
 import argparse
+import html.parser
 
-def fetch_profile(token, user):
+# This is a hacky HTML parser to extract profile information from GitHub
+class GHProfileParser(html.parser.HTMLParser):
+
+    @staticmethod
+    def has_itemprop(attrs, value):
+        for attr in attrs:
+            if attr[0] == 'itemprop' and attr[1] == value:
+                return True
+        return False
+
+    @staticmethod
+    def get_attr(attrs, key):
+        for attr in attrs:
+            if attr[0] == key:
+                return attr[1]
+        return None
+
+    def __init__(self):
+        super().__init__()
+        self.profile = {}
+        self.in_works_for = False
+        self.in_homeloc = False
+        self.in_url = False
+        self.in_name = False
+        self.in_followers = False
+
+    def set_once(self, key, value):
+        if key not in self.profile:
+            self.profile[key] = value
+
+    def handle_starttag(self, tag, attrs) -> None:
+        if tag == 'li' and self.has_itemprop(attrs, 'worksFor'):
+            self.in_works_for = True
+        if tag == 'li' and self.has_itemprop(attrs, 'homeLocation'):
+            self.in_homeloc = True
+        if tag == 'li' and self.has_itemprop(attrs, 'url'):
+            self.in_url = True
+        if tag == 'span' and self.has_itemprop(attrs, 'name'):
+            self.in_name = True
+        if tag == 'span' and len(attrs) == 1 and attrs[0][0] == 'class' and attrs[0][1] == 'text-bold color-fg-default':
+            self.in_followers = True
+        if tag == 'span' and self.in_works_for:
+            self.set_once('company', self.get_attr(attrs, 'title'))
+        if tag == 'a' and self.in_url:
+            self.set_once('blog', self.get_attr(attrs, 'href'))
+        if tag == 'a' and self.has_itemprop(attrs, 'image'):
+            self.set_once('avatar_url', self.get_attr(attrs, 'href'))
+        return super().handle_starttag(tag, attrs)
+
+    def handle_data(self, data: str) -> None:
+        if self.in_name:
+            self.set_once('name', data.strip())
+        if self.in_homeloc and data.strip():
+            self.set_once('location', data.strip())
+        if self.in_followers:
+            self.set_once('followers', data.strip())
+        return super().handle_data(data)
+
+def fetch_api_profile(token, user):
     url = "https://api.github.com/users/{}".format(user)
     headers = {
         'Accept': 'application/vnd.github+json',
@@ -15,29 +74,42 @@ def fetch_profile(token, user):
     }
     req = urllib.request.Request(url, None, headers)
     response = urllib.request.urlopen(req).read()
-    return json.loads(response)
+    profile = json.loads(response)
+    if profile['followers'] > 1000:
+        profile['followers'] = '{}k'.format(profile['followers'] // 1000)
+    else:
+        profile['followers'] = '{}'.format(profile['followers'])
+    return profile
+
+def fetch_html_profile(user):
+    url = "https://github.com/{}".format(user)
+    response = urllib.request.urlopen(url)
+    html_profile = response.read().decode('utf-8')
+    parser = GHProfileParser()
+    parser.feed(html_profile)
+    return parser.profile
 
+def tweak_profile(profile):
+    if 'location' in profile:
+        profile['location'] = profile['location'].replace(', ', ',')
+    profile['followers'] = profile['followers'] + ' followers'
+    return profile
 
 if __name__ == '__main__':
     parser = argparse.ArgumentParser(description='Creates a GitHub badge')
     parser.add_argument('username', type=str, help='GitHub username')
     parser.add_argument('-q', '--qrcode', action='store_true', help='Use QR code instead of profile image')
     args = parser.parse_args()
-    token = os.environ['GITHUB_TOKEN']
-    if not token:
-        print("Please set GITHUB_TOKEN environment variable")
-        sys.exit(1)
     username = args.username
-    profile = fetch_profile(token, username)
+    token = os.environ.get('GITHUB_TOKEN')
+    if not token:
+        print('GITHUB_TOKEN environment variable not set, using HTML parser')
+        profile = fetch_html_profile(username)
+    else:
+        profile = fetch_api_profile(token, username)
+    profile = tweak_profile(profile)
     profile_pic = profile['avatar_url']
     name = profile['name']
-    location = profile['location']
-    company = profile['company']
-    blog = profile['blog']
-    email = profile['email']
-    if location and len(location) > 15:
-        # try to make it shorter by removing some spaces
-        location = location.replace(', ', ',')
     tag = ggtag.GGTag()
     tag.rect(10, 25, 110, 110)
     if not args.qrcode:
@@ -55,16 +127,19 @@ def fetch_profile(token, user):
     else:
         tag.text(140, 50, 5, name)
         tag.text(140, 80, 2, "github.com/{}".format(username))
-    if location:
-        tag.icon(13, 156, 16, 'map-marker-alt')
-        tag.text(33, 158, 2, location)
-    if company:
-        tag.icon(13, 183, 16, 'building')
-        tag.text(33, 185, 2, company)
-    if blog:
-        tag.icon(180, 154, 16, 'link')
-        tag.text(202, 158, 2, blog)
-    if email:
-        tag.icon(180, 185, 16, 'envelope')
-        tag.text(202, 185, 2, email)
+
+    placeholder_icons = [(13, 156), (13, 183), (180, 154), (180, 185)]
+    placeholder_texts = [(33, 158), (33, 185), (202, 158), (202, 185)]
+    icon_map = {'location': 'map-marker-alt',
+                'company': 'building',
+                'blog': 'link',
+                'email': 'envelope',
+                'followers': 'user-friends'}
+    props = ['location', 'company', 'blog', 'followers', 'email']
+    index = 0
+    for prop in props:
+        if profile.get(prop) and index < 4:
+            tag.icon(*placeholder_icons[index], 16, icon_map[prop])
+            tag.text(*placeholder_texts[index], 2, profile[prop])
+            index += 1
     tag.browse()