diff options
Diffstat (limited to 'src')
-rwxr-xr-x | src/index-page.sh (renamed from src/index.sh) | 23 | ||||
-rwxr-xr-x | src/index-terms.sh | 38 | ||||
-rw-r--r-- | src/lib/create_table | 6 | ||||
-rw-r--r-- | src/lib/insert | 9 |
4 files changed, 60 insertions, 16 deletions
diff --git a/src/index.sh b/src/index-page.sh index 5aca876..e26cdf8 100755 --- a/src/index.sh +++ b/src/index-page.sh @@ -1,14 +1,13 @@ #!/usr/bin/env sh +# shellcheck source=./lib/create_table +. "$(dirname "$0")"/lib/create_table +# shellcheck source=./lib/insert +. "$(dirname "$0")"/lib/insert + DB_PATH="$1" shift -create_table() { - sqlite3 "$DB_PATH" "CREATE TABLE IF NOT EXISTS searchIndex(id INTEGER PRIMARY KEY, name TEXT, type TEXT, path TEXT);" - sqlite3 "$DB_PATH" "CREATE UNIQUE INDEX IF NOT EXISTS anchor ON searchIndex (name, type, path);" -} - - get_title() { FILE="$1" @@ -30,14 +29,6 @@ get_type() { esac } -insert() { - NAME="$1" - TYPE="$2" - PAGE_PATH="$3" - - sqlite3 "$DB_PATH" "INSERT INTO searchIndex(name, type, path) VALUES (\"$NAME\",\"$TYPE\",\"$PAGE_PATH\");" -} - insert_pages() { # Get title and insert into table for each html file while [ -n "$1" ]; do @@ -60,11 +51,11 @@ insert_pages() { esac if [ -n "$PAGE_NAME" ]; then - insert "$PAGE_NAME" "$PAGE_TYPE" "$(basename "$1")" + insert "$DB_PATH" "$PAGE_NAME" "$PAGE_TYPE" "$(basename "$1")" fi shift done } -create_table +create_table "$DB_PATH" insert_pages "$@" diff --git a/src/index-terms.sh b/src/index-terms.sh new file mode 100755 index 0000000..8ac3163 --- /dev/null +++ b/src/index-terms.sh @@ -0,0 +1,38 @@ +#!/usr/bin/env sh + +# shellcheck source=./lib/create_table +. "$(dirname "$0")"/lib/create_table +# shellcheck source=./lib/insert +. "$(dirname "$0")"/lib/insert + +TYPE="$1" +shift +DB_PATH="$1" +shift + +insert_index_terms() { + # Get each term from an index page and insert + while [ -n "$1" ]; do + grep -Eo "<a href.*</a>:" "$1" | while read -r line; do + insert_term "$line" + done + + shift + done +} + +insert_term() { + LINK="$1" + NAME="$(echo "$LINK" | pup -p 'a text{}' | sed 's/\"\"//g' | tr -d \\n)" + PAGE_PATH="$(echo "$LINK" | pup -p 'a attr{href}')" + + # DEBUG + #echo "$NAME" + #echo "$PAGE_PATH" + #echo + + insert "$DB_PATH" "$NAME" "$TYPE" "$PAGE_PATH" +} + +create_table "$DB_PATH" +insert_index_terms "$@" diff --git a/src/lib/create_table b/src/lib/create_table new file mode 100644 index 0000000..cdeb477 --- /dev/null +++ b/src/lib/create_table @@ -0,0 +1,6 @@ +create_table() { + DB_PATH="$1" + + sqlite3 "$DB_PATH" "CREATE TABLE IF NOT EXISTS searchIndex(id INTEGER PRIMARY KEY, name TEXT, type TEXT, path TEXT);" + sqlite3 "$DB_PATH" "CREATE UNIQUE INDEX IF NOT EXISTS anchor ON searchIndex (name, type, path);" +} diff --git a/src/lib/insert b/src/lib/insert new file mode 100644 index 0000000..2e4dd8c --- /dev/null +++ b/src/lib/insert @@ -0,0 +1,9 @@ +insert() { + DB_PATH="$1" + NAME="$2" + TYPE="$3" + PAGE_PATH="$4" + + sqlite3 "$DB_PATH" "INSERT INTO searchIndex(name, type, path) VALUES (\"$NAME\",\"$TYPE\",\"$PAGE_PATH\");" +} + |