https://github.com/dogsheep/pocket-to-sqlite/tree/0.2.3 and https://pypi.org/project/pocket-to-sqlite/0.2.3/

I tested this locally with python -m build and then pip install ...whl in a fresh virtual environment.

I think swarm-to-sqlite needs to avoid this error, maybe by setting up foreign keys in another way - or even by skipping foreign keys entirely on databases that don't support this kind of operation.

OK, I figured this out. Unfortunately it's an error that occurs on Python versions that have defensive mode turned on, and it doesn't look like there's a way to turn that mode off. See notes above.

Related: - https://github.com/simonw/sqlite-utils/issues/235

I just had that error myself on macOS while running the tests: ERROR tests/test_save_checkin.py::test_tables - sqlite3.OperationalError: table sqlite_master may not be modified ERROR tests/test_save_checkin.py::test_venue - sqlite3.OperationalError: table sqlite_master may not be modified ERROR tests/test_save_checkin.py::test_event - sqlite3.OperationalError: table sqlite_master may not be modified ERROR tests/test_save_checkin.py::test_sticker - sqlite3.OperationalError: table sqlite_master may not be modified ERROR tests/test_save_checkin.py::test_likes - sqlite3.OperationalError: table sqlite_master may not be modified ERROR tests/test_save_checkin.py::test_with_ - sqlite3.OperationalError: table sqlite_master may not be modified ERROR tests/test_save_checkin.py::test_users - sqlite3.OperationalError: table sqlite_master may not be modified ERROR tests/test_save_checkin.py::test_photos - sqlite3.OperationalError: table sqlite_master may not be modified ERROR tests/test_save_checkin.py::test_posts - sqlite3.OperationalError: table sqlite_master may not be modified ERROR tests/test_save_checkin.py::test_view - sqlite3.OperationalError: table sqlite_master may not be modified pytest --pdb shows it happening in the bit that adds foreign keys: ```

/Users/simon/.local/share/virtualenvs/swarm-to-sqlite-daPW7yIJ/lib/python3.9/site-packages/sqlite_utils/db.py(1096)add_foreign_keys() -> cursor.execute( (Pdb) list 1096 >> cursor.execute( 1097 "UPDATE sqlite_master SET sql = ? WHERE name = ?", 1098 (new_sql, table_name), 1099 ) 1100 cursor.execute("PRAGMA schema_version = %d" % (schema_version + 1)) 1101 -> cursor.execute("PRAGMA writable_schema = 0") 1102 # Have to VACUUM outside the transaction to ensure .foreign_keys property 1103 # can see the newly created foreign key. 1104 self.vacuum() ```

It's a rate limiting problem: https://github.com/dogsheep/github-to-sqlite/actions/runs/4622674009/jobs/8175633155

File "/home/runner/work/github-to-sqlite/github-to-sqlite/github_to_sqlite/cli.py", line 407, in commits utils.save_commits(db, commits, repo_full["id"]) File "/home/runner/work/github-to-sqlite/github-to-sqlite/github_to_sqlite/utils.py", line 635, in save_commits for commit in commits: File "/home/runner/work/github-to-sqlite/github-to-sqlite/github_to_sqlite/utils.py", line 417, in fetch_commits for commits in paginate(url, headers): File "/home/runner/work/github-to-sqlite/github-to-sqlite/github_to_sqlite/utils.py", line 478, in paginate raise GitHubError.from_response(response) github_to_sqlite.utils.GitHubError: ('API rate limit exceeded for user ID 9599.', 403) Error: Process completed with exit code 1.

I improved the readability by removing some unnecessary table aliases: sql with recursive nested_folders(folder_id, descendant_folder_id) as ( -- base case: select all immediate children of the root folder select id, id from folders where parent is null union all -- recursive case: select all children of the previous level of nested folders select nested_folders.folder_id, folders.id from nested_folders join folders on nested_folders.descendant_folder_id = folders.parent ) -- Find notes within all descendants of folder 1 select * from notes where folder in ( select descendant_folder_id from nested_folders where folder_id = 1 );

Here's a query that returns all notes in folder 1, including notes in descendant folders: sql with recursive nested_folders(folder_id, descendant_folder_id) as ( -- base case: select all immediate children of the root folder select id, id from folders where parent is null union all -- recursive case: select all children of the previous level of nested folders select nf.folder_id, f.id from nested_folders nf join folders f on nf.descendant_folder_id = f.parent ) -- Find notes within all descendants of folder 1 select * from notes where folder in ( select descendant_folder_id from nested_folders where folder_id = 1 ); With assistance from ChatGPT. Prompts were:

``` SQLite schema:

CREATE TABLE [folders] ( [id] INTEGER PRIMARY KEY, [long_id] TEXT, [name] TEXT, [parent] INTEGER, FOREIGN KEY([parent]) REFERENCES folders );

Write a recursive CTE that returns the following:

folder_id | descendant_folder_id

With a row for every nested child of every folder - so the top level folder has lots of rows Then I tweaked it a bit, then ran this: WITH RECURSIVE nested_folders(folder_id, descendant_folder_id) AS ( -- base case: select all immediate children of the root folder SELECT id, id FROM folders WHERE parent IS NULL UNION ALL -- recursive case: select all children of the previous level of nested folders SELECT nf.folder_id, f.id FROM nested_folders nf JOIN folders f ON nf.descendant_folder_id = f.parent ) -- select all rows from the recursive CTE SELECT * from notes where folder in (select descendant_folder_id FROM nested_folders where folder_id = 1)

Convert all SQL keywords to lower case, and re-indent ```

Here's what the CTE from that looks like:

My folders table will have:

• id - rowid
• long_id - that long unique string ID
• name - the name
• parent - foreign key to id

Improved script: zsh osascript -e 'tell application "Notes" set allFolders to folders repeat with aFolder in allFolders set folderId to id of aFolder set folderName to name of aFolder set folderContainer to container of aFolder if class of folderContainer is folder then set folderContainerId to id of folderContainer else set folderContainerId to "" end if log "ID: " & folderId log "Name: " & folderName log "Container: " & folderContainerId log " " end repeat end tell ' ``` ID: x-coredata://D2D50498-BBD1-4097-B122-D15ABD32BDEC/ICFolder/p6113 Name: Blog posts Container:

ID: x-coredata://D2D50498-BBD1-4097-B122-D15ABD32BDEC/ICFolder/p698 Name: JSK Container:

ID: x-coredata://D2D50498-BBD1-4097-B122-D15ABD32BDEC/ICFolder/p7995 Name: Nested inside blog posts Container: x-coredata://D2D50498-BBD1-4097-B122-D15ABD32BDEC/ICFolder/p6113

ID: x-coredata://D2D50498-BBD1-4097-B122-D15ABD32BDEC/ICFolder/p3526 Name: New Folder Container:

ID: x-coredata://D2D50498-BBD1-4097-B122-D15ABD32BDEC/ICFolder/p3839 Name: New Folder 1 Container:

ID: x-coredata://D2D50498-BBD1-4097-B122-D15ABD32BDEC/ICFolder/p2 Name: Notes Container:

ID: x-coredata://D2D50498-BBD1-4097-B122-D15ABD32BDEC/ICFolder/p6059 Name: Quick Notes Container:

ID: x-coredata://D2D50498-BBD1-4097-B122-D15ABD32BDEC/ICFolder/p7283 Name: UK Christmas 2022 Container: `` I filtered out things where the parent was an account and not a folder usingif class of folderContainer is folder then`.

Created through several rounds with ChatGPT (including hints like "rewrite that using setdefault()"): ```python def topological_sort(nodes): children = {} for node in nodes: parent_id = node["parent"] if parent_id is not None: children.setdefault(parent_id, []).append(node)

def traverse(node, result):
    result.append(node)
    if node["id"] in children:
        for child in children[node["id"]]:
            traverse(child, result)

sorted_data = []

for node in nodes:
    if node["parent"] is None:
        traverse(node, sorted_data)

return sorted_data

```

I used ChatGPT to write this: osascript -e 'tell application "Notes" set allFolders to folders repeat with aFolder in allFolders set folderId to id of aFolder set folderName to name of aFolder set folderContainer to container of aFolder set folderContainerName to name of folderContainer log "Folder ID: " & folderId log "Folder Name: " & folderName log "Folder Container: " & folderContainerName log " " --check for nested folders if count of folders of aFolder > 0 then set nestedFolders to folders of aFolder repeat with aNestedFolder in nestedFolders set nestedFolderId to id of aNestedFolder set nestedFolderName to name of aNestedFolder set nestedFolderContainer to container of aNestedFolder set nestedFolderContainerName to name of nestedFolderContainer log " Nested Folder ID: " & nestedFolderId log " Nested Folder Name: " & nestedFolderName log " Nested Folder Container: " & nestedFolderContainerName log " " end repeat end if end repeat end tell ' Which for my account output this: ``` Folder ID: x-coredata://D2D50498-BBD1-4097-B122-D15ABD32BDEC/ICFolder/p6113 Folder Name: Blog posts Folder Container: iCloud

Nested Folder ID: x-coredata://D2D50498-BBD1-4097-B122-D15ABD32BDEC/ICFolder/p7995
Nested Folder Name: Nested inside blog posts
Nested Folder Container: Blog posts

Folder ID: x-coredata://D2D50498-BBD1-4097-B122-D15ABD32BDEC/ICFolder/p698 Folder Name: JSK Folder Container: iCloud

Folder ID: x-coredata://D2D50498-BBD1-4097-B122-D15ABD32BDEC/ICFolder/p7995 Folder Name: Nested inside blog posts Folder Container: Blog posts

Folder ID: x-coredata://D2D50498-BBD1-4097-B122-D15ABD32BDEC/ICFolder/p3526 Folder Name: New Folder Folder Container: iCloud

Folder ID: x-coredata://D2D50498-BBD1-4097-B122-D15ABD32BDEC/ICFolder/p3839 Folder Name: New Folder 1 Folder Container: iCloud

Folder ID: x-coredata://D2D50498-BBD1-4097-B122-D15ABD32BDEC/ICFolder/p2 Folder Name: Notes Folder Container: iCloud

Folder ID: x-coredata://D2D50498-BBD1-4097-B122-D15ABD32BDEC/ICFolder/p6059 Folder Name: Quick Notes Folder Container: iCloud

Folder ID: x-coredata://D2D50498-BBD1-4097-B122-D15ABD32BDEC/ICFolder/p7283 Folder Name: UK Christmas 2022 Folder Container: iCloud ``` So I think the correct approach here is to run code at the start to list all of the folders (no need to do fancy recursion though, just a flat list with the parent containers is enough) and create a model of that hierarchy in SQLite.

Then when I import notes I can foreign key reference them back to their containing folder.

I'm tempted to use rowid for the foreign keys because the official IDs are pretty long.

So it looks like folders can be hierarchical?

From the Script Editor library docs:

A note has a:

• container (folder), r/o) : the folder of the note

Here's what a folder looks like:

folder n : a folder containing notes elements:

• contains folders, notes; contained by application, accounts, folders.

properties:

• name (text) : the name of the folder
• id (text, r/o) : the unique identifier of the folder
• shared (boolean, r/o) : Is the folder shared?
• container (account or folder, r/o) : the container of the folder

In terms of the UI: I'm tempted to say that the default behaviour is for it to run until it sees a note that it already knows about AND that has matching update/created dates, and then stop.

You can do a full import again ignoring that logic with apple-notes-to-sqlite notes.db --full.

It looks like the iteration order is most-recently-modified-first - I tried editing a note a bit further back in my notes app and it was the first one output by apple-notes-to-sqlite --dump.

OK, this works!

It doesn't have tests yet. I guess I'll need to mock subprocess to test this.

Before that conversion:

Monday, March 6, 2023 at 11:55:15 AM

After:

2023-03-06T11:55:15

Converting AppleScript date strings to ISO format is hard!

https://forum.latenightsw.com/t/formatting-dates/841 has a recipe I'll try:

set todayISO to (todayDate as «class isot» as string)

Not clear to me how timezones work here. I'm going to ignore them for the moment.

My prototype showed that images embedded in notes come out in the HTML export as bas64 image URLs, which is neat.

I opened the "Script Editor" app on my computer, used Window -> Library to open the Library panel, then clicked on the Notes app there. I got this:

So the notes object has these properties:

• name (text) : the name of the note (normally the first line of the body)
• id (text, r/o) : the unique identifier of the note
• container (folder, r/o) : the folder of the note
• body (text) : the HTML content of the note
• plaintext (text, r/o) : the plaintext content of the note
• creation date (date, r/o) : the creation date of the note
• modification date (date, r/o) : the modification date of the note
• password protected (boolean, r/o) : Is the note password protected?
• shared (boolean, r/o) : Is the note shared?

I'm going to ignore the concept of attachments for the moment.

I think the AppleScript I want to pass to osascript looks like this: applescript tell application "Notes" repeat with eachNote in every note set noteId to the id of eachNote set noteTitle to the name of eachNote set noteBody to the body of eachNote log "------------------------" & "\n" log noteId & "\n" log noteTitle & "\n\n" log noteBody & "\n" end repeat end tell But there are a few more properties I'd like to get - created and updated date for example.

This did the job! Next step is to turn that into a Python script.

After a few more rounds I got to this script, which outputs them to a /tmp/notes.txt file:

```zsh

!/bin/zsh

osascript -e ' set notesFile to "/tmp/notes.txt" set fileRef to open for access notesFile with write permission tell application "Notes" repeat with eachNote in every note set noteId to the id of eachNote set noteTitle to the name of eachNote set noteBody to the body of eachNote write "------------------------" & "\n" to fileRef write noteId & "\n" to fileRef write noteTitle & "\n\n" to fileRef write noteBody & "\n" to fileRef end repeat end tell close access fileRef' Then I wrote this little Python script to load them into a database:python import sqlite_utils split = b"------------------------\n" s = open("/tmp/notes.txt", "rb").read() notes = [n.decode("mac_roman") for n in s.split(split) if n]

cleaned_notes = [{ "id": n.split("\n")[0], "title": n.split("\n")[1], "body": "\n".join(n.split("\n")[2:]).strip() } for n in notes]

db = sqlite_utils.Database("/tmp/notes.db") db["notes"].insert_all(cleaned_notes) ```

Me:

Write applescript to loop through all of my Apple Notes and output their contents

ChatGPT:

applescript tell application "Notes" repeat with eachNote in every note set noteTitle to the name of eachNote set noteBody to the body of eachNote display dialog noteTitle & return & return & noteBody end repeat end tell Me:

How can I turn that into a zsh script which will output the results to my terminal

ChatGPT: ```zsh

!/usr/bin/zsh

osascript -e 'tell application "Notes" repeat with eachNote in every note set noteTitle to the name of eachNote set noteBody to the body of eachNote set output to noteTitle & "\n" & noteBody & "\n" display dialog output log output end repeat end tell' `` I had to change that first line to#!/bin/zsh` - then when I ran it it spammed my screen with dialog boxes.

Me:

I want to log the notes contents to the terminal, not display it in a dialog

ChatGPT:

```zsh

!/bin/zsh

osascript -e 'tell application "Notes" repeat with eachNote in every note set noteTitle to the name of eachNote set noteBody to the body of eachNote set output to noteTitle & "\n" & noteBody & "\n" log output end repeat end tell' ``` This really helps!

That was clearly the intention from the description of this issue: - #4

Agreed, that would be a much better implementation.

Just saw that's what's implemented here already! - #7

I thought this might need a documentation update but --all is already covered: https://github.com/dogsheep/pocket-to-sqlite/blob/0.2.1/README.md

Great catch, thanks.

I'm going to use it to mean --auth - since other tools in the Dogsheep family have the same convention.

--all will be the only way to specify all.

That fixed it.

Takes 30m to deploy the demo!

pycmarkgfm doesn't implement the Markdown plugin extension I was using.

I'm going to drop the GFM rendering from the demo, and just treat it as regular markdown.

Here's how the demo is deployed: https://github.com/dogsheep/github-to-sqlite/blob/dbac2e5dd8a562b45d8255a265859cf8020ca22a/.github/workflows/deploy-demo.yml#L103-L119

I'm suspicious of py-gfm, which is used like this:

https://github.com/dogsheep/github-to-sqlite/blob/dbac2e5dd8a562b45d8255a265859cf8020ca22a/demo-metadata.json#L49-L51

https://github.com/Zopieux/py-gfm says that library is no longer maintained, and suggests https://github.com/Zopieux/pycmarkgfm as an alternative.

Thanks!

That's interesting - yeah it looks like the number of pages can be derived from the Link header, which is enough information to show a progress bar, probably using Click just to avoid adding another dependency.

https://docs.github.com/en/rest/guides/traversing-with-pagination

First attempt at this: sql select 'issues' as "table", id, node_id, title, user, created_at, body, repo from issues union all select 'issue_comments' as "table", issue_comments.id, issue_comments.node_id, '' as title, issue_comments.user, issue_comments.created_at, issue_comments.body, issues.repo from issue_comments join issues on issues.id = issue_comments.issue order by created_at desc https://github-to-sqlite.dogsheep.net/github?sql=select+%27issues%27+as+%22table%22%2C+id%2C+node_id%2C+title%2C+user%2C+created_at%2C+body%2C+repo%0D%0Afrom+issues%0D%0Aunion+all%0D%0Aselect+%27issue_comments%27+as+%22table%22%2C+issue_comments.id%2C+issue_comments.node_id%2C+%27%27+as+title%2C+issue_comments.user%2C+issue_comments.created_at%2C+issue_comments.body%2C+issues.repo%0D%0Afrom+issue_comments+join+issues+on+issues.id+%3D+issue_comments.issue%0D%0Aorder+by+created_at+desc

Thanks for this!

I tried this ampersand fix: https://regex101.com/r/ojU2H9/1 ```python

https://regex101.com/r/ojU2H9/1

_invalid_ampersand_re = re.compile(r'&(?![a-z0-9]+;)')

def fix_bad_xml(xml): # More fixes for things like '&' not as part of an entity return _invalid_ampersand_re.sub('&', xml) ```

Even with that I'm still getting total garbage in the <en-note> content - it's just HTML, not even trying to be XML.

It looks like I was using the round-trip to dump the <?xml version="1.0" encoding="UTF-8" standalone="no"?> and <!DOCTYPE prefixes.

https://github.com/dogsheep/evernote-to-sqlite/blob/36a466f142e5bad52719851c2fbda0c05cd35b99/evernote_to_sqlite/utils.py#L34-L42

Not sure why I was round-tripping the content_xml like that - I will try not doing that.

The debugger showed me that it broke on a string that looked like this: ```xml

<en-note>

Q3 2018 Reflection & Development

... ``` Yeah that is not valid XML!

Also on workout_points.workout_id to speed up queries to show all points in a specific workout.

Workaround:

sqlite-utils create-index healthkit.db workout_points -- -date

See https://sqlite-utils.datasette.io/en/stable/cli.html#creating-indexes

Thanks!

Maybe Click changed something which meant that this broke things when it didn't used to?

This looks like the bug to me:

https://github.com/dogsheep/twitter-to-sqlite/blob/197e69cec40052c423a5ed071feb5f7cccea41b9/twitter_to_sqlite/cli.py#L239-L241

type=str, default=False

Weird, added debug code and got this: {'screen_name': 'simonw', 'count': 200, 'since_id': 'False', 'tweet_mode': 'extended'} - so maybe it's a twitter-to-sqlite bug where somehow the string False is being passed somewhere.

https://developer.twitter.com/en/docs/twitter-api/v1/tweets/timelines/api-reference/get-statuses-user_timeline says the API has been replaced by the new v2 one, but it should still work - and the since_id parameter is still documented on that page.

I'm getting this too. Looking into it now.

Got a TIL out of this: https://til.simonwillison.net/jq/extracting-objects-recursively

If you hit the endpoint for a comment that's part of a thread you get that comment and its recursive children: https://hn.algolia.com/api/v1/items/27941552

You can tell that it's not the top-level because the parent_id isn't null. You can use story_id to figure out what the top-level item is.

json { "id": 27941552, "created_at": "2021-07-24T15:08:39.000Z", "created_at_i": 1627139319, "type": "comment", "author": "nine_k", "title": null, "url": null, "text": "<p>I wish ...", "points": null, "parent_id": 27941108, "story_id": 27941108 }

Prototype:

curl 'https://hn.algolia.com/api/v1/items/27941108' \
  | jq '[recurse(.children[]) | del(.children)]' \
  | sqlite-utils insert hn.db items - --pk id

I'm going to build a general-purpose hacker-new-to-sqlite search ... command, where one of the options is to search within the URL.

Prototype:

curl "https://hn.algolia.com/api/v1/search_by_date?query=simonwillison.net&restrictSearchableAttributes=url&hitsPerPage=1000" | \
  jq .hits | sqlite-utils insert hn.db items - --pk objectID --alter

https://hn.algolia.com/api/v1/search_by_date?query=simonwillison.net&restrictSearchableAttributes=url looks like it does what I want.

https://hn.algolia.com/api/v1/search_by_date?query=simonwillison.net&restrictSearchableAttributes=url&hitsPerPage=1000 - returns 1000 at once.

Otherwise you have to paginate using &page=2 etc - up to nbPages pages.

https://www.algolia.com/doc/api-reference/api-parameters/hitsPerPage/ says 1000 is the maximum.

That page doesn't have an API but does look easy to scrape.

The other option here is the HN Search API powered by Algolia, documented at https://hn.algolia.com/api

Unfortunately I don't think updating the database is practical, because the export doesn't include unique identifiers which can be used to update existing records and create new ones. Recreating from scratch works around that limitation.

I've not explored workouts with SpatiaLite but that's a really good idea.

I'm definitely interested in supporting events in this tool - see #14.

Have you found a way to access events in GraphQL? I can only see way to access a timeline of events for a single issue or a single pull request. See also https://github.community/t/get-event-equivalent-for-v4/13600/2

Thanks for catching this.

Thanks!

I'm not sure why but my most recent import, when displayed in Datasette, looks like this:

Sorting by id in the opposite order gives me the data I would expect - so it looks like a bunch of null/blank messages are being imported at some point and showing up first due to ID ordering.

I imported my 10GB mbox with 750,000 emails in it, ran this tool (with a hacked fix for the blob column problem) - and now a search that returns 92 results takes 25.37ms! This is fantastic.

Solution could be to pre-process that string by splitting on ( and dropping everything afterwards, assuming that the (...) bit isn't necessary for correctly parsing the date.

I added this code to output a message ID on errors: diff print("Errors: {}".format(num_errors)) print(traceback.format_exc()) + print("Message-Id: {}".format(email.get("Message-Id", "None"))) continue Having found a message ID that had an error, I ran this command to see the context:

rg --text --context 20 '44F289B0.000001.02100@SCHWARZE-DWFXMI' ~/gmail.mbox

This was for the following error: File "/Users/simon/Dropbox/Development/google-takeout-to-sqlite/google_takeout_to_sqlite/utils.py", line 102, in get_mbox message["date"] = get_message_date(email.get("Date"), email.get_from()) File "/Users/simon/Dropbox/Development/google-takeout-to-sqlite/google_takeout_to_sqlite/utils.py", line 178, in get_message_date datetime_tuple = email.utils.parsedate_tz(mail_date) File "/Library/Developer/CommandLineTools/Library/Frameworks/Python3.framework/Versions/3.7/lib/python3.7/email/_parseaddr.py", line 50, in parsedate_tz res = _parsedate_tz(data) File "/Library/Developer/CommandLineTools/Library/Frameworks/Python3.framework/Versions/3.7/lib/python3.7/email/_parseaddr.py", line 69, in _parsedate_tz data = data.split() AttributeError: 'Header' object has no attribute 'split' Here's what I spotted in the ripgrep output: 177133570:Message-Id: <44F289B0.000001.02100@SCHWARZE-DWFXMI> 177133571-Date: Mon, 28 Aug 2006 08:14:08 +0200 (Westeurop�ische Sommerzeit) 177133572-X-Mailer: IncrediMail (5002253) So it could it be that _parsedate_tz is having trouble with that Mon, 28 Aug 2006 08:14:08 +0200 (Westeurop�ische Sommerzeit) string.

3 also mentions the conflicting version with other tools.

Looks like you're doing this: python elif message.get_content_type() == "text/plain": body = message.get_payload(decode=True) So presumably that decodes to a unicode string?

I imagine the reason the column is a BLOB for me is that sqlite-utils determines the column type based on the first batch of items - https://github.com/simonw/sqlite-utils/blob/09c3386f55f766b135b6a1c00295646c4ae29bec/sqlite_utils/db.py#L1927-L1928 - and I got unlucky and had something in my first batch that wasn't a unicode string.

Confirmed: removing the len() call does not speed things up, so it's reading through the entire file for some other purpose too.

It looks like the body is being loaded into a BLOB column - so in Datasette default it looks like this:

If I datasette install datasette-render-binary and then try again I get this:

It would be great if we could store the body as unicode text instead. May have to do something clever to decode it based on some kind of charset header?

I got 9 warnings that look like this: Errors: 1 Traceback (most recent call last): File "/Users/simon/Dropbox/Development/google-takeout-to-sqlite/google_takeout_to_sqlite/utils.py", line 103, in get_mbox message["date"] = get_message_date(email.get("Date"), email.get_from()) File "/Users/simon/Dropbox/Development/google-takeout-to-sqlite/google_takeout_to_sqlite/utils.py", line 167, in get_message_date datetime_tuple = email.utils.parsedate_tz(mail_date) File "/Library/Developer/CommandLineTools/Library/Frameworks/Python3.framework/Versions/3.7/lib/python3.7/email/_parseaddr.py", line 50, in parsedate_tz res = _parsedate_tz(data) File "/Library/Developer/CommandLineTools/Library/Frameworks/Python3.framework/Versions/3.7/lib/python3.7/email/_parseaddr.py", line 69, in _parsedate_tz data = data.split() AttributeError: 'Header' object has no attribute 'split' It would be useful if those warnings told me the message ID (or similar) of the affected message so I could grep for it in the mbox and see what was going on.

I'm not sure if it would work, but there is an alternative pattern for showing a progress bar against a really large file that I've used in healthkit-to-sqlite - you set the progress bar size to the size of the file in bytes, then update a counter as you read the file.

https://github.com/dogsheep/healthkit-to-sqlite/blob/3eb2b06bfe3b4faaf10e9cf9dfcb28e3d16c14ff/healthkit_to_sqlite/cli.py#L24-L57 and https://github.com/dogsheep/healthkit-to-sqlite/blob/3eb2b06bfe3b4faaf10e9cf9dfcb28e3d16c14ff/healthkit_to_sqlite/utils.py#L4-L19 (the progress_callback() bit) is where that happens.

It can be a bit of a convoluted pattern, and I'm not at all sure it would work for mbox files since it looks like that library has other reasons it needs to do a file scan rather than streaming it through one chunk of bytes at a time. So I imagine this would not work here.

The command takes quite a while to start running, presumably because this line causes it to have to scan the WHOLE file in order to generate a count:

https://github.com/dogsheep/google-takeout-to-sqlite/blob/a3de045eba0fae4b309da21aa3119102b0efc576/google_takeout_to_sqlite/utils.py#L66-L67

I'm fine with waiting though. It's not like this is a command people run every day - and without that count we can't show a progress bar, which seems pretty important for a process that takes this long.

The Rich-powered progress bar is pretty:

Wow, my mbox is a 10.35 GB download!

Thanks!

I requested my Gmail export from takeout - once that arrives I'll test it against this and then merge the PR.

I've merged these changes by hand now, thanks!

My best guess is that you have an older version of sqlite-utils installed here - the replace=True argument was added in version 2.0. I've bumped the dependency in setup.py.

Aha! I think I've figured out what's going on here.

The CData blocks containing the notes look like this:

<![CDATA[<!DOCTYPE en-note SYSTEM "http://xml.evernote.com/pub/enml2.dtd"><en-note><div>This note includes two images.</div><div><br /></div>...

The DTD at http://xml.evernote.com/pub/enml2.dtd includes some entities:

```

%HTMLlat1;

%HTMLsymbol;

%HTMLspecial; `` So I need to be able to handle all of those different entities. I think I can do that usinghtml.entities.entitydefs` from the Python standard library, which looks a bit like this:

```python {'Aacute': 'Á', 'aacute': 'á', 'Aacute;': 'Á', 'aacute;': 'á', 'Abreve;': 'Ă', 'abreve;': 'ă', 'ac;': '∾', 'acd;': '∿',

...

} ```

Thanks for the fix!

I really like the way you're using pipes here - really smart. It's similar to how I build the demo database in this GitHub Actions workflow:

https://github.com/dogsheep/github-to-sqlite/blob/62dfd3bc4014b108200001ef4bc746feb6f33b45/.github/workflows/deploy-demo.yml#L52-L82

twitter-to-sqlite actually has a mechanism for doing this kind of thing, documented at https://github.com/dogsheep/twitter-to-sqlite#providing-input-from-a-sql-query-with---sql-and---attach

It lets you do things like:

$ twitter-to-sqlite users-lookup my.db --sql="select follower_id from following" --ids Maybe I should add something similar to github-to-sqlite? Feels like it could be really useful.

This is interesting - how did you create that initial table? Was this using the twitter-to-sqlite import archive.db ~/Downloads/twitter-2019-06-25-b31f2.zip command, or something else?

I just got a similar error:

``` File "/home/dogsheep/datasette-venv/lib/python3.8/site-packages/swarm_to_sqlite/utils.py", line 79, in save_checkin checkins_table.m2m("users", user, m2m_table="with", pk="id") File "/home/dogsheep/datasette-venv/lib/python3.8/site-packages/sqlite_utils/db.py", line 2048, in m2m id = other_table.insert(record, pk=pk, replace=True).last_pk File "/home/dogsheep/datasette-venv/lib/python3.8/site-packages/sqlite_utils/db.py", line 1781, in insert return self.insert_all( File "/home/dogsheep/datasette-venv/lib/python3.8/site-packages/sqlite_utils/db.py", line 1899, in insert_all self.insert_chunk( File "/home/dogsheep/datasette-venv/lib/python3.8/site-packages/sqlite_utils/db.py", line 1709, in insert_chunk result = self.db.execute(query, params) File "/home/dogsheep/datasette-venv/lib/python3.8/site-packages/sqlite_utils/db.py", line 226, in execute return self.conn.execute(sql, parameters) pysqlite3.dbapi2.OperationalError: table users has no column named countryCode

```

Here's why:

if "fts5" in str(e):

But the error being raised here is:

sqlite3.OperationalError: no such column: to

I'm going to attempt the escaped on on every error.

Looks like I already do that here: https://github.com/dogsheep/dogsheep-beta/blob/9ba4401017ac24ffa3bc1db38e0910ea49de7616/dogsheep_beta/init.py#L141-L146

One fix for this could be to try running the raw query, but if it throws an error run it again with the query escaped.

I deliberately added support for advanced FTS in https://github.com/dogsheep/dogsheep-beta/commit/cbb2491b85d7ff416d6d429b60109e6c2d6d50b9 for #13 but that's the cause of this bug.

The memory profiler tricks I used in https://github.com/dogsheep/healthkit-to-sqlite/issues/7 could help figure out what's going on here.

I'm just going to add q for the moment.

Should I pass any other details to the display_sql here as well?

To do this I'll need the search term to be passed to the display_sql SQL query: https://github.com/dogsheep/dogsheep-beta/blob/4890ec87b5e2ec48940f32c9ad1f5aae25c75a4d/dogsheep_beta/init.py#L164-L171

I think I can do this as a cunning trick in display_sql. Consider this example query: https://til.simonwillison.net/tils?sql=select%0D%0A++path%2C%0D%0A++snippet%28til_fts%2C+-1%2C+%27b4de2a49c8%27%2C+%278c94a2ed4b%27%2C+%27...%27%2C+60%29+as+snippet%0D%0Afrom%0D%0A++til%0D%0A++join+til_fts+on+til.rowid+%3D+til_fts.rowid%0D%0Awhere%0D%0A++til_fts+match+escape_fts%28%3Aq%29%0D%0A++and+path+%3D+%27asgi_lifespan-test-httpx.md%27%0D%0A&q=pytest

sql select path, snippet(til_fts, -1, 'b4de2a49c8', '8c94a2ed4b', '...', 60) as snippet from til join til_fts on til.rowid = til_fts.rowid where til_fts match escape_fts(:q) and path = 'asgi_lifespan-test-httpx.md' The and path = 'asgi_lifespan-test-httpx.md' bit means we only get back a specific document - but the snippet highlighting is applied to it.

I don't want to add a full HTML parser (like BeautifulSoup) as a dependency for this feature. Since the HTML comes from a single, trusted source (GitHub) I could probably handle this using regular expressions.

I'm going to rewrite those <a href="#filtering-tables"> links to <a href="#user-content-filtering-tables"> - but only if a corresponding id="user-content-filtering-tables" element exists.

Thanks for this!

Given rate limits (see #51) this command might be better implemented by running a git clone into a temporary directory - doing so would retrieve all of the files in one go.