Make htmlify_tweet() more robust when encountering strs not ints

Some tweet JSON have the `display_text_range` set as strings not ints,
e.g. `["0", "140"]` rather than `[0, 140]`. Particularly when the
JSON has come from the downloaded Twitter archive.

And

Some tweet JSON have the `["entities"][<kind>]["indices"]` set as
strings not ints, e.g. `["0", "9"]` rather than `[0, 9]`.
Particularly when the JSON has come from the downloaded Twitter
archive.

For #229

ditto/twitter/utils.py

@@ -74,6 +74,21 @@ def htmlify_tweet(json_data):
     if "entities" in json_data and "symbols" not in json_data["entities"]:
         json_data["entities"]["symbols"] = []
 
+    # Some Tweets (eg from a downloaded archive) have strings instead of ints
+    # to define text ranges. ["0", "140"] rather than [0, 140].
+    # We fix those here so that Twython doesn't complain.
+    if "display_text_range" in json_data:
+        json_data["display_text_range"] = [
+            int(n) for n in json_data["display_text_range"]
+        ]
+    if "entities" in json_data:
+        for key, value in json_data["entities"].items():
+            for count, entity in enumerate(value):
+                if "indices" in entity:
+                    json_data["entities"][key][count]["indices"] = [
+                        int(n) for n in entity["indices"]
+                    ]
+
     # This does most of the work for us:
     # https://twython.readthedocs.org/en/latest/usage/special_functions.html#html-for-tweet
     html = Twython.html_for_tweet(

tests/twitter/fixtures/api/tweet_with_display_text_range_str.json

@@ -0,0 +1,20 @@
+{
+  "retweeted": false,
+  "source": "<a href=\"http://tapbots.com/tweetbot\" rel=\"nofollow\">Tweetbot for iΟS</a>",
+  "entities": {
+    "hashtags": [],
+    "symbols": [],
+    "user_mentions": [],
+    "urls": []
+  },
+  "display_text_range": ["0", "140"],
+  "favorite_count": "6",
+  "id_str": "915152022273449987",
+  "truncated": false,
+  "retweet_count": "0",
+  "id": "915152022273449987",
+  "created_at": "Tue Oct 03 09:50:19 +0000 2017",
+  "favorited": false,
+  "full_text": "Open iPad Slack, scroll to bottom. Close Slack. Open Slack; it’s scrolled back up. Every time. Maddening. Not biggest problem in world, BUT.",
+  "lang": "en"
+}

tests/twitter/fixtures/api/tweet_with_entities_indices_str.json

@@ -0,0 +1,40 @@
+{
+  "retweeted": false,
+  "source": "<a href=\"https://tapbots.com/software/tweetbot/mac\" rel=\"nofollow\">Tweetbot for Mac</a>",
+  "entities": {
+    "hashtags": [],
+    "symbols": [],
+    "user_mentions": [
+      {
+        "name": "Terry Collier",
+        "screen_name": "terrycol",
+        "indices": ["0", "9"],
+        "id_str": "123",
+        "id": "123"
+      },
+      {
+        "name": "Bob Ferris",
+        "screen_name": "bobferris",
+        "indices": ["10", "20"],
+        "id_str": "234",
+        "id": "234"
+      }
+    ],
+    "urls": []
+  },
+  "display_text_range": [0, 142],
+  "favorite_count": "0",
+  "in_reply_to_status_id_str": "914906397061664768",
+  "id_str": "914908752482111488",
+  "in_reply_to_user_id": "123",
+  "truncated": false,
+  "retweet_count": "0",
+  "id": "914908752482111488",
+  "in_reply_to_status_id": "914906397061664768",
+  "created_at": "Mon Oct 02 17:43:39 +0000 2017",
+  "favorited": false,
+  "full_text": "@terrycol @bobferris I liked it and only thought some of it was a bit silly. But analysis beyond that is probably beyond the scope of Twitter.",
+  "lang": "en",
+  "in_reply_to_screen_name": "terrycol",
+  "in_reply_to_user_id_str": "123"
+}

tests/twitter/test_utils.py

@@ -35,6 +35,7 @@ def test_htmlify_description(self):
         )
 
 
+
 class HtmlifyTweetEntitiesTestCase(HtmlifyTestCase):
     "Linkify URLs from entities: urls, screen_names, hashtags."
 
@@ -160,6 +161,49 @@ def test_strip(self):
         self.assertEqual("Y", tweet_html[-1])
 
 
+class HtmlifyTweetStrsNotIntsTestCase(HtmlifyTestCase):
+
+    def test_handles_display_text_range_str(self):
+        """Cope correctly if display_text_range is strings, not ints.
+
+        Some tweet JSON have the display_text_range set as strings not ints,
+        e.g. ["0", "140"] rather than [0, 140]. Particularly when the
+        JSON has come from the downloaded Twitter archive.
+        It should be able to cope with that.
+        """
+        api_fixture = "tweet_with_display_text_range_str.json"
+        tweet_html = htmlify_tweet(self.getJson(api_fixture))
+        self.assertEqual(
+            tweet_html,
+            (
+                "Open iPad Slack, scroll to bottom. Close Slack. Open Slack; it’s "
+                "scrolled back up. Every time. Maddening. Not biggest problem in "
+                "world, BUT."
+            )
+        )
+
+    def test_handles_entities_indicies_str(self):
+        """Cope correctly if entities' indicies are strings, not ints.
+
+        Some tweet JSON have the ["entities"][<kind>]["indices"] set as
+        strings not ints, e.g. ["0", "9"] rather than [0, 9].
+        Particularly when the JSON has come from the downloaded Twitter
+        archive.
+        It should be able to cope with that.
+        """
+        api_fixture = "tweet_with_entities_indices_str.json"
+        tweet_html = htmlify_tweet(self.getJson(api_fixture))
+        self.assertEqual(
+            tweet_html,
+            (
+                '<a href="https://twitter.com/terrycol" rel="external">@terrycol</a> '
+                '<a href="https://twitter.com/bobferris" rel="external">@bobferris</a> '
+                'I liked it and only thought some of it was a bit silly. But analysis '
+                'beyond that is probably beyond the scope of Twitter.'
+            )
+        )
+
+
 class HtmlifyTweetUrlsTestCase(HtmlifyTestCase):
     "Further tests for specific problems with URLs."