mySpindel/LittleBock/extract.nim

41 lines
1.2 KiB
Nim
Raw Normal View History

2022-05-06 17:33:17 +02:00
import std/os
import std/strutils
import std/htmlparser
import std/xmltree
import std/json
import std/strtabs # To access XmlAttributes
# extract iSpindel JSON from little bock HTML files
proc jsparse(data: string): JsonNode =
# get JSON tree
var tree: JsonNode = parseJson(data)
if tree.kind == JObject and tree.hasKey("series"):
tree = tree["series"]
for branch in tree:
for unwanted in ["color","opacity","yAxis","dashStyle"]:
if branch.hasKey(unwanted): branch.delete(unwanted)
return tree
proc hparse(hfile: string): string =
# parse a named HTML document
let html: XmlNode = loadHtml(hfile)
for hdiv in html.findAll("div"):
if hdiv.attrs.hasKey("id"):
if hdiv.attrs["id"] == "fermentation_log_chart":
let data: string = hdiv.attrs["data-chart-options"]
return pretty(jsparse(data))
proc jsextract(hfile: string): string =
# run the parsers
result = "None"
var js: string = hparse(hfile)
if not isEmptyOrWhitespace(js):
let jsfile: string = hfile.changeFileExt("json")
result = jsfile
writeFile(jsfile, js)
for hfile in walkFiles("*.html"):
echo hfile & " >> " & jsextract(hfile)