[This is preliminary documentation and subject to change.]
Here is an excerpt from a dump file:
1. Chunk ID: ........... 2
2. Chunk Break Type: ... END OF SENTENCE
3. Chunk State: ........ TEXT
4. Chunk Locale: ....... 0x411
5. Chunk Source ID: .... 2
6. Chunk Start Source .. 0x0
7. Chunk Length Source . 0x0
8. GUID ................ b725f130-47ef-101a-a5f1-02608c9eebac
9. Property ID ......... 0x13
10. This is a HTML filter test page
11. Chunk ID: ........... 3
12. Chunk Break Type: ... END OF SENTENCE
13. Chunk State: ........ TEXT
14. Chunk Locale: ....... 0x411
15. Chunk Source ID: .... 2
16. Chunk Start Source .. 0x0
17. Chunk Length Source . 0x0
18. GUID ................ f29f85e0-4ff9-1068-ab91-08002b27b3d9
19. Property ID ......... 0x2
20. This is a HTML filter test page
21. Chunk ID: ........... 4
22. Chunk Break Type: ... END OF SENTENCE
23. Chunk State: ........ VALUE
24. Chunk Locale: ....... 0x411
25. Chunk Source ID: .... 2
26. Chunk Start Source .. 0x0
27. Chunk Length Source . 0x0
28. GUID ................ f29f85e0-4ff9-1068-ab91-08002b27b3d9
29. Property ID ......... 0x2
30. This is a HTML filter test page
The first 9 lines describe the current chunk structure. Note that the GUID and the Property ID correspond to PSGUID_STORAGE / PID_STG_CONTENTS. This is a chunk containing plain text. The text follows the chunk structure: "This is a HTML filter test page".
The next chunk, starting at line (11), has a different GUID, corresponding to the HTML filter, and a different propid, corresponding to a HTML Href. This is a pseudo-property, exported by the HTML filter. The next chunk, starting at line (21), has the same GUID and propid, but its chunk state is VALUE instead of TEXT. Note that the text in these last two chunks are the same as for the first chunk. The filter wanted three attributes (plain text, HTML Href as text, and HTML Href as a value) to be applied to this particular phrase, so it was emitted in three separate chunks.