diff options
author | Vincent Ambo <mail@tazj.in> | 2023-01-21T15·00+0300 |
---|---|---|
committer | tazjin <tazjin@tvl.su> | 2023-01-21T17·49+0000 |
commit | 8eeb5d3bccf831681b2cad5c3b322e6a08f596df (patch) | |
tree | 4e02105c8ab39a4466b29e98c16aa9e2c69ac9b4 /corp/russian/data-import/src/db_setup.rs | |
parent | 429c0d00c4cd07ea90c85bf1ec2f2c742d970420 (diff) |
feat(corp/data-import): add import of OR 'words_forms' table r/5730
This is the full morphological set table for all the words from the lemmata table, which they don't call it that. Change-Id: I6f5be673c5f59f11e36bd8c8c935844a7d4fd170 Reviewed-on: https://cl.tvl.fyi/c/depot/+/7894 Tested-by: BuildkiteCI Reviewed-by: tazjin <tazjin@tvl.su>
Diffstat (limited to 'corp/russian/data-import/src/db_setup.rs')
-rw-r--r-- | corp/russian/data-import/src/db_setup.rs | 39 |
1 files changed, 38 insertions, 1 deletions
diff --git a/corp/russian/data-import/src/db_setup.rs b/corp/russian/data-import/src/db_setup.rs index 5fe64717ad..4644edf094 100644 --- a/corp/russian/data-import/src/db_setup.rs +++ b/corp/russian/data-import/src/db_setup.rs @@ -6,7 +6,7 @@ //! introduce things like foreign key constraints between tables that //! represent relations. -use super::{bail, Ensure}; +use super::Ensure; use crate::oc_parser::*; use crate::or_parser; use log::{debug, info}; @@ -181,6 +181,16 @@ CREATE TABLE or_words ( word_type TEXT, level TEXT ) STRICT; + +CREATE TABLE or_words_forms ( + id INTEGER PRIMARY KEY, + word_id INTEGER NOT NULL, + form_type TEXT, + position TEXT, + form TEXT, + form_bare TEXT, + FOREIGN KEY(word_id) REFERENCES words(id) +) STRICT; "#, ) .ensure("setting up OpenRussian table schema failed"); @@ -215,3 +225,30 @@ VALUES (?1, ?2, ?3, ?4, ?5, ?6, ?7) info!("inserted {} OpenRussian words", count); } + +pub fn insert_or_word_forms<I: Iterator<Item = or_parser::WordForm>>(conn: &Connection, forms: I) { + let mut stmt = conn + .prepare_cached( + " +INSERT INTO or_words_forms (id, word_id, form_type, position, form, form_bare) +VALUES (?1, ?2, ?3, ?4, ?5, ?6) +", + ) + .ensure("failed to prepare OR word forms statement"); + let mut count = 0; + + for form in forms { + stmt.execute(( + form.id, + form.word_id, + form.form_type, + form.position, + form.form, + form.form_bare, + )) + .ensure("failed to insert OR word form"); + count += 1; + } + + info!("inserted {} OpenRussian word forms", count); +} |