ABOUT THIS ARCHIVE
========================================
Format: eFleckt Archive, version 3.

This folder is a copy of an eFleckt archive. Everything in it is plain
text, standard audio (.m4a) and JSON, so it can be read, searched and built
on without eFleckt.


THE FILES
----------------------------------------
entries/     One text file per story: title, date, chapter, the people,
             places and emotions in it, the summary, and the transcript,
             word for word. When audio is included, the recording sits
             beside it under the same name (.m4a); recordings added to a
             story later end in _part2, _part3. An interview may also have
             a _consent recording: the other person, or a parent, agreeing
             to be recorded.
cast/        The people in the stories.
places/      The places in the stories.
chapters/    The chapters of the life, in order, with their summaries.
notes/       What has been noted about each person, place and chapter.
about-me/    The life story and reflections eFleckt wrote from the stories.
photos/      The profile photo and photos of people.
data.json    Everything above as data, and more. Described below.


DATA.JSON
----------------------------------------
One object. Most keys are lists of rows, and rows point at each other by id.

entries         One row per story.
                  title, summary, transcript
                  type            collection, recollection, interview,
                                  question_of_the_day, character_study,
                                  location_study
                  captured_at     when it was recorded, on the clock of the
                                  phone that recorded it. The +00:00 is not
                                  a real time zone.
                  anchor_type, anchor_raw, anchor_week, estimated_period
                                  when the story is about, which can be
                                  years before it was told. anchor_raw is
                                  how it was said ("the summer I was
                                  seven"); anchor_week is the week of the
                                  owner's life, counting from 0 at birth.
                  visibility      standard or private
                  chapter_id      -> chapters.id
                  deleted_at      set on a story that was deleted and is
                                  still inside its recovery window. It has
                                  no text file.
                  speaker_segments  on an interview: who spoke, turn by turn
                  export_text_file, export_audio_file,
                  export_consent_audio_file
                                  where this story is in this folder

entryAudio      One row per recording. entry_id -> entries.id.
                  segment_order   0 is the story's first recording; higher
                                  numbers were added later
                  transcript      what was said in this recording
                  duration_seconds  missing on older recordings
                  export_audio_file  where it is in this folder

chapters        name, sort_order, start_date, summary, is_current.
                  start_week      as anchor_week: the week of life it begins

castMembers     People. name, relationship, aliases, summary.
                  status          confirmed, mentioned, dismissed, declined.
                                  The last two are people the owner said
                                  no to; they are left out of cast/.

locations       Places. name, type, city, region, country, latitude,
                longitude, summary. status as for people.

entryCast       Which people are in which story (entry_id, cast_member_id).

entryLocations  Which places are in which story (entry_id, location_id).
                  relevance       primary or mentioned

entryEmotions   One row per emotion read in a story. entry_id -> entries.id.
                  emotions.name, emotions.category
                  score           0 to 1
                  source          llm = read from the words;
                                  hume, gemini_audio = read from the voice;
                                  blended = both

entityNotes     Things noted about a person, place, chapter or the owner.
                  entity_type + entity_id say which one
                  source          entry_extraction = drawn from a story;
                                  manual = written by the owner

userSummaries   About Me. section_key says which piece; is_current marks
                the version shown in the app.

agentQuestions, castQuestions, savedPrompts
                Questions eFleckt asked, and ones that were saved.

profile, credits, reminders, consentRecords, legacyGrants
                Account records. Not in an archive passed to a legacy
                contact.

_meta           format, version, and counts: the number of rows in each
                list, so this file can be checked against the account.


WHAT IS LEFT OUT
----------------------------------------
- The number vectors eFleckt uses for search.
- Internal processing fields, payment-provider ids, and the tokens behind
  legacy invitations.
- From the readable files only: deleted stories, and people and places the
  owner said no to. data.json still lists them, marked as above.
- From an archive passed to a legacy contact: Private stories, deleted
  stories, people and places the owner said no to, and the account records.
