From 3a12056d5f5ea2686935d2bc484e91be2ee24c92 Mon Sep 17 00:00:00 2001 From: Peter Kleiweg Date: Sat, 23 May 2026 13:47:20 +0200 Subject: [PATCH] opschonen www --- www/README | 3 - www/app.html | 104 ++++- www/app.js | 16 + www/gettags.sh | 57 --- www/mkAll.py | 70 ---- www/nieuwe-namen.html | 673 ------------------------------- www/style.css | 151 ++++--- www/style2.css | 146 ------- www/template.html | 21 - www/top2html.py | 65 --- www/weekreeks.html | 899 ------------------------------------------ 11 files changed, 221 insertions(+), 1984 deletions(-) delete mode 100644 www/README delete mode 100755 www/gettags.sh delete mode 100755 www/mkAll.py delete mode 100644 www/nieuwe-namen.html delete mode 100644 www/style2.css delete mode 100644 www/template.html delete mode 100755 www/top2html.py delete mode 100644 www/weekreeks.html diff --git a/www/README b/www/README deleted file mode 100644 index c47dbb9..0000000 --- a/www/README +++ /dev/null @@ -1,3 +0,0 @@ -for i in ../data/*2026-14-4*; do ./top2html.py $i > `basename $i`.table; done -for i in *t20*; do rm -f `basename $i .t20.table`.table; done - diff --git a/www/app.html b/www/app.html index 2c56d1e..df1d3c6 100644 --- a/www/app.html +++ b/www/app.html @@ -5,7 +5,7 @@ - + @@ -86,5 +86,107 @@

+
+

Bronnen

+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
Algemeen
NieuwsNL
NOS
NU
Reporters Online
Sargasso
Amsterdam
AT5
+ In de buurt Amsterdam +
+ Parool Amsterdam +
Groningen
+ Gemeente Groningen +
+ In de buurt Groningen +
Oog
RTV Noord
Sikkom
Literatuur
+ Literair Nederland +
Tzum
Vlaanderen
HLN
VRT NWS
+
diff --git a/www/app.js b/www/app.js index 5a0ee90..da033ca 100644 --- a/www/app.js +++ b/www/app.js @@ -1,5 +1,6 @@ var dates var datesNr +var countsWeek var parts = [ 'nieuwe namen', @@ -29,6 +30,18 @@ function sleep(ms) { return new Promise((resolve) => setTimeout(resolve, ms)) } +function setCounts(week) { + if (week != countsWeek) { + countsWeek = week + var max = data[week].max + var pp = Object.entries(data[week].sources) + for (var i in pp) { + document.getElementById(pp[i][0]).style.width = + (pp[i][1] / max) * 100 + '%' + } + } +} + function getJSON(url) { return new Promise(function (resolve, reject) { var xhr = new XMLHttpRequest() @@ -136,6 +149,7 @@ async function loadSource(source, week) { tab.appendChild(tr) d.appendChild(tab) idData.innerHTML = d.innerHTML + setCounts(week) } async function loadPart(part, week) { @@ -156,6 +170,7 @@ async function loadPart(part, week) { tab.appendChild(tr) d.appendChild(tab) idData.innerHTML = d.innerHTML + setCounts(week) } async function loadWeken(source, part) { @@ -180,6 +195,7 @@ async function loadWeken(source, part) { tab.appendChild(tr) d.appendChild(tab) idData.innerHTML = d.innerHTML + setCounts(dates[datesNr].week) } function locateWeek(date) { diff --git a/www/gettags.sh b/www/gettags.sh deleted file mode 100755 index 43acb62..0000000 --- a/www/gettags.sh +++ /dev/null @@ -1,57 +0,0 @@ -#!/bin/bash - -unset CDPATH -PATH=/net/corpora/nlnieuws/bin:/net/aps/bin:$PATH -export TZ=Europe/Amsterdam - -item=tag -if [ "$1" = "cat" ] -then - item=cat - shift -fi - -if [ "$1" = "" ] -then - ds=`date -d -7days +%G-%V` -else - case "$1" in - 2[0-9][0-9][0-9]-[0-5][0-9]) - ds=$1 - ;; - *) - echo INVALID - exit 1 - ;; - esac -fi - -w=4 - -cd /net/corpora/nlnieuws - -collect() { - declare -gA counts - declare -i val - count="$1" - shift - word="$*" - val=${counts["$word"]} - counts["$word"]=$(( $val + $count )) -} - -for file in $( find . -name $ds-$w.$item.txt -or $( week2files $ds $w | sed -e "s/data.dz/$item.txt/g" ) ) -do - IFS=$'\n' - for line in $(cat $file) - do - IFS=' '$'\t'$'\n' - collect $line - done -done - -for i in "${!counts[@]}" -do - printf "%8d\t%s\n" ${counts[$i]} "$i" - #echo -e "${counts[$i]}\t$i" -done | sort -f -k 2 | sort -n -r -k 1,1 -s diff --git a/www/mkAll.py b/www/mkAll.py deleted file mode 100755 index 61bd191..0000000 --- a/www/mkAll.py +++ /dev/null @@ -1,70 +0,0 @@ -#!/usr/bin/env python3 - -import sys -import re -import subprocess - -head = ''' - - - {} {} week {} - - - - - - - - -
-

{} {} week {}

- terugkijkend over vier weken -
-
-''' - -tail='''
- - -''' - -namen = { - 'algemeen': 'Algemeen', - 'VRT': 'Vlaanderen', - 'groningen': 'Groningen', - 'AT5': 'Amsterdam', - 'Tzum': 'Literatuur' -} - -verbose=False -if sys.argv[1] == '-v': - verbose=True - sys.argv = sys.argv[:1] + sys.argv[2:] - -ep=sys.argv[1] -if not re.match('^2[0-9][0-9][0-9]-[0-5][0-9]$', ep): - print("Ongeldig patroon '", ep, "', moet yyyy-ww zijn") - sys.exit(1) - -jaar=ep[:4] -week=ep[5:].lstrip('0') - -for base in ('algemeen', 'VRT', 'groningen', 'AT5', 'Tzum'): - name = namen[base] - with open(name + '.html', 'wt', encoding='utf-8') as fp: - fp.write(head.format(name, jaar, week, name, jaar, week)) - fp.flush() - for part in ('nieuwe-namen', 'nieuwe-woorden', 'personen', 'overige-namen', 'locaties', 'organisaties'): - if verbose: - print(base, part) - if part == 'locaties': - fp.write('\n