A2 Loading a family tree (GEDCOM)

The chapter says what the app takes over from a GEDCOM file from another program, what it skips and how it admits this in the import summary. Describes the Strom app 3.21.2 (and the research 1.25.2); changes from a later version carry its number. Opening the research's file and linking a tree to the research belong to chapter S1. The whole path of a family tree from another program (reliability marks, protection of the family tree's facts, the card What next, export back) is described in S6.

A2.1 What is loaded and what is not

The app takes over the people, families, sources, notes, events and media it knows. What it does not know it does not carry over, but it admits it: each tag not carried over is counted in the tile Unsupported and named in the row Skipped records. Nothing is lost silently at the level of a record or of a direct fact of a person, family and source.

• Limits: unknown records (level 0) and unknown direct facts of a person, family and source (level 1) are counted. Unknown details deeper in a record (level 2 and below) are skipped without counting. Bookkeeping tags of other programs (_UPD, RIN, _UID, CHAN, _PROJECT_GUID, _EXPORTED_FROM_SITE_ID) are skipped silently; they say nothing about the family. The tags ANCI, DESI, RFN, AFN and RESN are counted. A number of facts without a place of their own in the app are not thrown away but written into the person's note as a labelled line: blessing, retirement, social standing, description, identification number, number of children, number of marriages, property, insurance number, a general fact and "Also known as". Strom's research blocks are not read from an ordinary file and are counted (see A2.7).
• Controls: Import → Import GEDCOM (a .ged file). In a new tree also the button on the empty tree. The summary see A2.1.1.
• Since: app 1.9.0 (summary of tags not carried over); note lines from app 3.0.0.
• From app 3.26.0:
  • _UID, RIN, AFN and other program IDs (UID, EXID, _FSFTID) are not skipped: they are kept with the person and the family together with the original record identifier (@I12@) as program IDs (S6.1);
  • the program's own tags (_XXX) directly on a person, family and source are kept for the export Back to {program} (S6.6); deeper own tags are counted;
  • MyHeritage photo cutouts (_POSITION, _CUTOUT) are bookkeeping and are not counted;
  • a divorce written before the marriage (PAF) is not lost and does not overwrite the marriage; an empty marriage or divorce written before another one with a date (MyHeritage) keeps the date and place of the dated one;
  • a name written as a description (Son, Unknown) comes back as it was written, with no note made of it; text on the line of a fact (education, burial, a custom event), an address in parts (ADR1, CITY…) and a web link in the text of a source stay and come back as the file had them.

A2.1.1 Import summary

After the conversion the window GEDCOM conversion opens. The tiles People, Partnerships, Placeholders and Unsupported are always there; Photos, Documents, Excerpts, Sources and Events only with a count above zero. The row Skipped records lists the tags from the most frequent ("OBJE ×12"). Further rows report people with unknown sex and children recorded in more than one family. "Everything in the file was imported." is shown only when nothing was skipped. Sex is not guessed from the role in the family; a value other than M or F is unknown sex.

Options: Import as a new tree (the current tree stays), Merge with existing (see A2.8), Save as JSON; from an empty tree also Insert into tree. The checkbox Import images (excerpts, attachments, photos) shows an estimate of the size. Its default comes from Settings → Load images when importing and holds only for this import. Without images, transcriptions, archives and page references stay. Paths and addresses of files are offered for linking.

• From app 3.22.0: above the list of skipped tags there is a sentence saying what the tile Unsupported counts: parts of the file that Strom does not know and did not carry over, not people and relationships.
• From app 3.26.0:
  • the line “Kept for the export “Back to {program}”: N tags (…)” when the file has the program's own tags;
  • the row Text encoding for a file with a guessed or single-byte encoding (A2.2);
  • after the import of another program's file, the card What next (S6.4); it replaces the earlier questions after an import;
  • Insert into tree first saves a backup of the tree (S6.5).

A2.2 Encoding and GEDCOM version

The app reads GEDCOM 5.5.1 and GEDCOM 7 and recognizes the encoding by itself. It always exports GEDCOM 5.5.1 in UTF-8.

• Limits: the encoding is set first by a BOM mark (UTF-8, UTF-16 LE and BE). Without one the line 1 CHAR in the header decides: ANSEL has its own conversion (an unknown byte shows as "�"), UNICODE is UTF-16, ANSI is Windows-1252, anything else UTF-8. Line ends CR, LF and CRLF are equal. The continuations CONC and CONT are joined; for places, names, pages, addresses and notes a space at the start of a continuation is kept too. From GEDCOM 7, SNOTE is read as a shared note, and records without an identifier are read. Standalone media records 0 @O1@ OBJE are not read and are counted in Unsupported; an image referenced only by a pointer therefore does not arrive. A GEDZIP package is not loaded; the window accepts only .ged.
• Controls: none; the encoding is not chosen.
• Since: app 1.16.0 (ANSEL conversion and encoding by the header).
• From app 3.22.0: the GEDCOM 7 tags UID, CREA and EXID are skipped silently as bookkeeping data; in 3.21.2 they are counted in Unsupported.
• From app 3.26.0:
  • CHAR CP1250 or WINDOWS-1250 is Windows-1250. ANSI means the code page of the computer that wrote the file: valid UTF-8 stays UTF-8, otherwise the words of the file decide between Windows-1250 (Central Europe: č ř ě ł ż) and Windows-1252 (Western Europe). A file without CHAR and a file with CHAR UTF-8 that is not valid UTF-8 are guessed the same way;
  • when the encoding was guessed or is single-byte and the file has text with diacritics, the import summary shows the row Text encoding with Names: and a choice of another encoding (UTF-8, Windows-1250 (Central Europe), Windows-1252 (Western Europe), ISO-8859-2 (Latin-2), ANSEL (older GEDCOM)). Garbled letters in the names (è instead of č, ø instead of ř) mean that another encoding fits. Controls: the choice in the row Text encoding;
  • standalone media records 0 @O1@ OBJE are read through the references to them; a family photo referenced this way is only counted.
• From app 3.27.0: a family's photo (1 OBJE at a family, written in place or as a reference @O1@) is not shown in the app, but is kept as the file wrote it for the export Back to {program} (S6.6). A media record the family refers to goes back as a photo written directly at the family. The import summary lists it in the row “Kept for the export “Back to {program}”…”, not in Unsupported. Not for a research file.

A2.3 Given names and surnames

The given name and surname are taken from the subtags GIVN and SURN, otherwise from the line NAME by the slashes. Titles from NPFX and NSFX are kept separately and removed from the line; without these tags a title is not guessed. The rules are the same as in the research.

• Limits: without slashes the line is split at the first space. Text after the closing slash belongs to the surname. Entries such as "N.N.", "neznámý", "unknown", "nieznany" give the name "?". Descriptions instead of a name ("syn", "dcera", "mrtvě narozený", "totgeboren", "filius", text in brackets) give "?" and the words go into the person's note. Further NAME lines are variants. A name with TYPE married or _MARNM is the married surname, not a variant; the last such one wins. When the married name comes first and the birth name only later, they are swapped. 2 _STROM_PRIMARY Y marks the name the person goes by. A person without a given name and surname is a placeholder person (tile Placeholders). Groups of surname variants and family lines are covered by A4.
• Controls: none; the result is edited in the person's card.
• Since: app 3.10.0 (titles), 3.13.0 (married surname), 3.18.0 (_STROM_PRIMARY).

A2.4 Dates and their forms

GEDCOM dates are converted to the app's form: 3 JUN 1900 to a day, JUN 1900 to a month, 1900 to a year. An estimate (ABT, EST, CAL, INT) is "about", BEF "before", AFT "after", BET … AND and FROM … TO a range. What does not fit the form stays in the person's note; nothing is thrown away.

• Limits: a double year (1699/00) keeps the first year. A calendar other than the Gregorian is removed and the note says "Date in the Julian calendar: …" (also Hebrew, French revolutionary, Roman). Text in brackets is kept as "Date as written: …". An unknown month leaves only the year, with a note. Words in Czech, Slovak, Polish, German, Russian and Ukrainian ("po", "před", "kolem", "mezi … a", "nach", "około") are read as tags only when a valid date follows them; "po válce" stays text. The age at an event (AGE) is written as "61 years", INFANT as "infant". A form of date that the app would write differently (e.g. EST or CAL) is remembered at the person and the relationship. The export returns it in its original form as long as the date does not change.
• Controls: none.
• Since: app 3.0.0 (notes on dates), 3.7.0 (words of languages), 3.17.0 (kept forms).
• From app 3.27.0: a month and year in figures (07/1987, 7/1987, 07.1987, 7.1987) are read as that month, also with a mark or a word before them (ABT, “cca”, “kolem”, “před”, “po”). Only a month 1–12 counts: 13/1987 is not read as a month. A double year (1699/00) and a date with a day (12.03.1919) are not read this way. Dates written the way people write them (cca 1850, 14/6/1850, mezi 1850-1852) are read as in A2.4.1.

A2.4.1 Dates written the way people write them

A date written the way people write it, and not in GEDCOM words, is read as a date. This holds alike for the date fields (person, event, union), for the GEDCOM import and for text that is already in the tree's data (e.g. from a JSON edited by hand).

• The words are read in Czech and also in English, German and Polish (“about”, “um”, “około”, “zwischen”, “między” …).
• A day and month that can be swapped (7/8/1850: both up to 12 and different) are read by the language of the app: in Czech, German and Polish the day first, in English the month first. The same GEDCOM file loaded in another language of the app thus gives a different date. An unambiguous date (25/3/1919, 3/3/1919) is read the same everywhere.
• The GEDCOM import, for a day and month that can be swapped and for “cca” before a range, leaves a note next to the date with the original notation (“Date as written: Cca 1850-1852”). Mezi 1850 - 1852 is read as the range 1850–1852.
• Text in the data is shown, sorted, ages are counted from it and it goes to GEDCOM as a date (ABT, BEF, AFT, BET … AND). The text itself does not change; saving the form of the person, event or union writes it in the app's notation.
• Limits:
  • a slash between two years (1850/1851) is read by the date field as a period from 1850 to 1851; the GEDCOM import leaves it as text (two years as an alternative);
  • text remains: a question mark (1850?), unknown digits (18XX), a quarter (Q4/1850), a season, “nebo”, “cca po” and “cca před” and values in words. The date field refuses them as an invalid date; the import leaves them in the note “Date as written: …” (it reads Cca po 1850 and Cca podzim 1850 as “about 1850” with a note);
  • a date that does not exist (31/4/1850), a year with five digits and a range whose start is after its end also remain text;
  • trees imported earlier are not corrected by themselves; a new GEDCOM import corrects them.
• Controls: the date field in the form; the GEDCOM import (A2.1).
• Since: app 3.27.0

A2.5 Relationships and the status of a couple

A family (FAM) gives the relationship of two partners and their children. The status of the couple is set by marriages and divorces in the order of the file; dated events are compared by date. The tag _STAT takes precedence over events.

• Limits: _STAT knows the values married, divorced, separated and unmarried partners (in English married, divorced, separated, partners, unmarried, cohabiting etc.). A divorce and a new marriage of the same couple give a note "Divorce …, new marriage …". The same couple in two families is joined. A third partner in one family is not carried over and is counted. A family with one parent gets a placeholder second parent "?". A family without children and without a parent stays only when it says something (a marriage, a date, a source, a note). A child recorded in more than one family is shown in the birth family; the others are written into a note ("Also recorded as an adopted child of …"). Adoption and fostering are read from PEDI and more precisely from _FREL and _MREL. The export writes _STAT only for unmarried and separated couples.
• Controls: none; the status is edited in the relationship in the person's card.
• Since: app 3.0.0 (_STAT), 1.16.0 (a child in more than one family), 3.8.0 (further events of a couple).
• From app 3.22.0: _RELATIONSHIP_STATUS is read too; in 3.21.2 it is counted in Unsupported. The values of both tags are recognized in Czech, Polish and German too ("manželé", "rozvedeni", "druh a družka", "rozwiedzeni", "Lebensgemeinschaft"). An unknown value is counted in Unsupported. When a family has both tags, _STAT wins.

A2.6 Sources, citations and QUAY

A source record (SOUR) gives a source of the app: title, archive, call number, transcription (TEXT), link and excerpts. A citation at a birth and a death proves the person's birth and death, a citation at a person the person, a citation at a family the couple. The rating QUAY is shown as Source reliability: 3 is Original, 2 Copy / index, 1 and 0 Uncertain.

• Limits: without TITL the abbreviation ABBR is taken, otherwise "?". Publisher, author and the call number in the archive go into the source's note. A citation with a page (PAGE) other than the source's creates a derived source for that page with the same title and archive; transcription and excerpts do not pass to it. From app 3.26.0 a citation of the same page with another record date (DATA › DATE) does the same. The reliability belongs to the citation (A2.6.1). DATA and TEXT inside a citation are not read. Every image at a source is an excerpt. The app does not change certainty from another program. How the research takes QUAY over as the form and information of a source is said by R3.2.1; when a fact is proven, by R3.8. Sources in the app are described in A3.
• Controls: Source reliability in the source's detail (Original, Copy / index, Uncertain).
• Since: app 1.9.4 (QUAY), 3.9.0 (derived source for a page).
• From app 3.22.0: the person's story (_STORY) also reads and exports references to sources.
• From app 3.27.0: the call number from REPO › CALN goes into the field Call number (the first CALN), not into the note; the publisher and the author stay in the note. The export writes the book as 1 REPO with 2 CALN and one readable PAGE composed of the fields: folio, other location, image ("fol. 12v, no. 5, image 34"; the word for the image in the app's language). A number the other location already names is not repeated. A source without a folio and an image is exported exactly as before. The source's fields are described in A3.9.

A2.6.1 Reliability at the citation

The same entry is often an original for one fact and secondary information for another: a census gives the occupation first hand, the age second hand. So QUAY belongs to the citation, not to the source. A source has a default reliability, and a citation with another value keeps its own.

• Enforced by the code:
  • citations of the same entry with the same page, the same record date and the same other lines of the citation that differ only in QUAY are citations of one source. The source is in Sources once, with its images (OBJE);
  • another page or another record date still makes a source of its own for that entry (A2.6);
  • the source's default reliability is the QUAY of its first citation (a citation without QUAY takes the record's QUAY when it has one); a citation with another QUAY keeps it, a citation without QUAY at a source with a default reliability stays without one;
  • the export (to the research, Back to {program} and To share with others) writes each citation's QUAY: its own, otherwise the source's default. A tree without citations with their own reliability is exported exactly as before. Loading and exporting again therefore changes the QUAY of no citation.
• A tree saved by an older version: on opening, the sources that loading in app 3.26.0 made only for another QUAY of the same page are joined back into their source. Their citations are moved to that source and the reliability passes to the citation. Before that the app saves the backup Before the data update of a new version (S6.5). Only a source that differs from its own only in reliability is joined; a source with its own note, transcript, excerpt or other text stays. The research the tree is connected to gets no change from this (the source and the page are the same), and the tag Image only in the research at the joined source disappears (A3.8).
• Limits: the reliability of a single citation is only shown in the app (A3.6, S7.1); only the default Source reliability can be changed, in the source editor (A3.1).
• Since: app 3.27.0 (in 3.26.0 a citation with another QUAY made a second source without images)

A2.7 Strom's own tags and what comes back

The app's export writes its own tags with the prefix _STROM, so that what GEDCOM 5.5.1 cannot hold comes back on a new load. The header 1 SOUR STROM says that Strom wrote the file.

• Limits: on export and a new load these come back: the married surname (as a second NAME with TYPE married), the name the person goes by, family lines, groups of surname variants, the person's questions and stories, the state of facts, the kind and fingerprint of images, the verification of a transcription (_STROM_VERIFIED), the reading of a record, excerpts with cropping, families without a couple (_STROM_NO_COUPLE), a child's relationship to the parents, and witnesses. These do not come back from an export: who read a transcription (_STROM_READ), an image left out because of its size, and place records. Research blocks (conflicts, hypotheses, searches, links, islands, drafts) are read only from the research's file, and the app's export does not write them. A file without Strom's header is read without its conventions: descriptions instead of a name go into the note (A2.3). The lines in the header that link to the research (_STROM_TREE etc.) are covered by S1 and S2.
• Controls: Export → Export GEDCOM.
• Since: app 3.9.0 or earlier (single tags were added up to 3.19.0).
• From app 3.26.0: program IDs come back under _STROM_IMPORT and _STROM_XID; on a new load the file is paired by them (S6.7). A file with Strom's header is not a family tree from another program: it has no mark From family tree and no card What next.
• From app 3.27.0: a source's folio, image and other location come back under _STROM_FOLIO, _STROM_IMAGE and _STROM_LOCATOR beside the composed PAGE; with the tags the fields hold, not PAGE. The export Back to {program} (S6.6) for the program the tree came from does not write them. The app sends them to the research only when the research says it reads them.

A2.8 Repeated import and merging

Each import as a new tree creates a new tree; the old trees do not change. Merge with existing pairs the people from the file with the people of the current tree and saves the result as yet another new tree (A8).

• Limits: a repeated import of the research's file into the research's tree is covered by S1.
• Controls: Import → Import GEDCOM → Import as a new tree or Merge with existing.
• Since: app 1.0.0
• From app 3.26.0: a family tree of the same program loaded again is also paired by program ID (reason Same ID in the file); a match of 50–79 is only offered (S6.7).