From 8addd2d58ab16bc13e2d822c2297181e20aafe78 Mon Sep 17 00:00:00 2001
From: Dima Gerasimov <karlicoss@gmail.com>
Date: Mon, 25 Sep 2023 10:41:36 +0100
Subject: [PATCH] new module: Harmonic app for Hackernews

---
 my/config.py              |   4 ++
 my/hackernews/common.py   |  18 ++++++
 my/hackernews/harmonic.py | 115 ++++++++++++++++++++++++++++++++++++++
 tox.ini                   |   1 +
 4 files changed, 138 insertions(+)
 create mode 100644 my/hackernews/harmonic.py

diff --git a/my/config.py b/my/config.py
index a59eadd..9cc9c11 100644
--- a/my/config.py
+++ b/my/config.py
@@ -265,3 +265,7 @@ class whatsapp:
     class android:
         export_path: Paths
         my_user_id: Optional[str]
+
+
+class harmonic:
+    export_path: Paths
diff --git a/my/hackernews/common.py b/my/hackernews/common.py
index 8c7dd1e..0c5ff9b 100644
--- a/my/hackernews/common.py
+++ b/my/hackernews/common.py
@@ -1,2 +1,20 @@
+from typing import Protocol
+
+from my.core import datetime_aware, Json
+
+
 def hackernews_link(id: str) -> str:
     return f'https://news.ycombinator.com/item?id={id}'
+
+
+class SavedBase(Protocol):
+    @property
+    def when(self) -> datetime_aware: ...
+    @property
+    def uid(self) -> str: ...
+    @property
+    def url(self) -> str: ...
+    @property
+    def title(self) -> str: ...
+    @property
+    def hackernews_link(self) -> str: ...
diff --git a/my/hackernews/harmonic.py b/my/hackernews/harmonic.py
new file mode 100644
index 0000000..a4eb28e
--- /dev/null
+++ b/my/hackernews/harmonic.py
@@ -0,0 +1,115 @@
+"""
+[[https://play.google.com/store/apps/details?id=com.simon.harmonichackernews][Harmonic]] app for Hackernews
+"""
+REQUIRES = ['lxml']
+
+from dataclasses import dataclass
+from datetime import datetime, timezone
+import json
+import html
+from pathlib import Path
+from typing import Any, Dict, Iterator, List, Optional, Sequence, TypedDict, cast
+
+from lxml import etree
+from more_itertools import unique_everseen, one
+
+from my.core import (
+    Paths,
+    Res,
+    Stats,
+    datetime_aware,
+    get_files,
+    stat,
+)
+from .common import hackernews_link, SavedBase
+
+from my.config import harmonic as user_config
+
+
+@dataclass
+class harmonic(user_config):
+    export_path: Paths
+
+
+def inputs() -> Sequence[Path]:
+    return get_files(harmonic.export_path)
+
+
+class Cached(TypedDict):
+    author: str
+    created_at_i: int
+    id: str
+    points: int
+    test: Optional[str]
+    title: str
+    type: str  # TODO Literal['story', 'comment']? comments are only in 'children' field tho
+    url: str
+    # TODO also has children with comments, but not sure I need it?
+
+
+# TODO reuse savedbase in materialistic?
+@dataclass
+class Saved(SavedBase):
+    raw: Cached
+
+    @property
+    def when(self) -> datetime_aware:
+        ts = self.raw['created_at_i']
+        return datetime.fromtimestamp(ts, tz=timezone.utc)
+
+    @property
+    def uid(self) -> str:
+        return self.raw['id']
+
+    @property
+    def url(self) -> str:
+        return self.raw['url']
+
+    @property
+    def title(self) -> str:
+        return self.raw['title']
+
+    @property
+    def hackernews_link(self) -> str:
+        return hackernews_link(self.uid)
+
+
+_PREFIX = 'com.simon.harmonichackernews.KEY_SHARED_PREFERENCES'
+
+
+def _saved() -> Iterator[Res[Saved]]:
+    for p in inputs():
+        # TODO defensive for each item!
+        tr = etree.parse(p)
+
+        res = one(cast(List[Any], tr.xpath(f'//*[@name="{_PREFIX}_CACHED_STORIES_STRINGS"]')))
+        cached_ids = [x.text.split('-')[0] for x in res]
+
+        cached: Dict[str, Cached] = {}
+        for sid in cached_ids:
+            res = one(cast(List[Any], tr.xpath(f'//*[@name="{_PREFIX}_CACHED_STORY{sid}"]')))
+            j = json.loads(html.unescape(res.text))
+            cached[sid] = j
+
+        res = one(cast(List[Any], tr.xpath(f'//*[@name="{_PREFIX}_BOOKMARKS"]')))
+        for x in res.text.split('-'):
+            ids, item_timestamp = x.split('q')
+            # not sure if timestamp is any useful?
+
+            cc = cached.get(ids, None)
+            if cc is None:
+                # TODO warn or error?
+                continue
+
+            yield Saved(cc)
+
+
+def saved() -> Iterator[Res[Saved]]:
+    yield from unique_everseen(_saved())
+
+
+def stats() -> Stats:
+    return {
+        **stat(inputs),
+        **stat(saved),
+    }
diff --git a/tox.ini b/tox.ini
index 9ec80f1..9487ae3 100644
--- a/tox.ini
+++ b/tox.ini
@@ -133,6 +133,7 @@ commands =
         my.github.ghexport        \
         my.goodreads              \
         my.google.takeout.parser  \
+        my.hackernews.harmonic    \
         my.hypothesis             \
         my.instapaper             \
         my.ip.all                 \