Consolidate placeholder tracking in the source tracker.

[ianjosephwilson]

• Morph SourceTracker into an Iterator.
• Move placeholder injection (and tracking) internally into the SourceTracker.
• Redirect TemplateParser's placeholder extraction to proxy methods of the `SourceTracker. The parser still needs to know when its just looking at a snapshot versus when it is "marking" the placeholder as found. Not ideal but this composition feels better and more distinct from the regular parsing duties.
• Use configure_source_tracker factory function because later we bind the PlaceholderConfig to members other than PlaceholderState that depend on the size of the placeholders (defined by the config).

The SourceTracker methods get_expression and format_starttag will be removed in another PR and those responsibilities will be in the SourceReader (constructed and accessed during exceptions).

[ianjosephwilson]

@davepeck Thanks for merging those changes. Here is the next "chunk" that continues to move things into alignment with the error messaging.

[davepeck]

LGTM! Two very small and ignorable comments:

1. There are a couple places in parser.py where we do if self.source: or if self.source and ...:. Maybe these should become if self.source is not None for future safety (it's fine now, but what happens if we add, say, __len__?)

2. Should we have a variant of test_iter that tests t"" and t"one-string-only" cases?

As for find_placeholders and remove_placeholders: I think it works, but agree you have to be cautious.

Anyway, extremely minor feedback! LMK when you're good for a merge.

[ianjosephwilson]

@davepeck I added a parametrized test for those simple string cases you mentioned. If this looks ok I think you should merge it so I can start prep'ing the next part.

[davepeck]

@ianjosephwilson Merged; thanks!

By the way, one thing I've been contemplating is moving the source structure to something more standard in the python world, specifically:

src/tdom/... <-- top level src/ which seems to be the thing to do these days
tests/... <-- top level tests/ that mirror the structure of src/tdom/ aka src/tdom/sub/whippersnapper.py gets tests/sub/test_whippersnapper.py

This is just a bunch of churn, so I want to avoid it until you are in a position to take it. If that's now, cool; if that's, like, after the error handling stuff is settled, cool too -- just let me know. (And if you don't like this structure, LMK what you'd prefer!)

[ianjosephwilson]

@ianjosephwilson Merged; thanks!

By the way, one thing I've been contemplating is moving the source structure to something more standard in the python world, specifically:

src/tdom/... <-- top level src/ which seems to be the thing to do these days tests/... <-- top level tests/ that mirror the structure of src/tdom/ aka src/tdom/sub/whippersnapper.py gets tests/sub/test_whippersnapper.py

This is just a bunch of churn, so I want to avoid it until you are in a position to take it. If that's now, cool; if that's, like, after the error handling stuff is settled, cool too -- just let me know. (And if you don't like this structure, LMK what you'd prefer!)

Haha don't you dare DAVE! Thanks for asking and I think these are both good ideas for once we get the error stuff in assuming nothing else is going on. But it would probably be safer to wait in case I need to scissor-and-glue-stick together more sub-sub PRs for us to digest to get this thing in.

[davepeck]

But it would probably be safer to wait

Yup, no rush on this one. Will do later.