jyapayne/Nim
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Better doc search (#8260)

Browse source 
* Modified the doc generation to produce a custom data attribute to allow for better search functionality

* Implemented fuzzy matching for the Nim Doc search instead of the simple regex match.

* Fix to the WordBoundry state transition from code review with @Varriount. Also removed silly testing template that is no longer used.

* Update fuzzysearch.nim

* Update fuzzysearch.nim

* Update fuzzysearch.nim

* Update dochack.nim

* Update dochack.nim
This commit is contained in:
Ray Imber 2018-07-20 02:58:42 -07:00 • committed by Varriount
commit 060871e64a
3 changed files with 167 additions and 20 deletions
Download patch file Download diff file Expand all files Collapse all files

36
tools/dochack/dochack.nim
View file

@ -1,6 +1,5 @@
import karax
import fuzzysearch
proc findNodeWith(x: Element; tag, content: cstring): Element =
if x.nodeName == tag and x.textContent == content:
@ -88,11 +87,11 @@ proc toHtml(x: TocEntry; isRoot=false): Element =
if ul.len != 0: result.add ul
if result.len == 0: result = nil
proc containsWord(a, b: cstring): bool {.asmNoStackFrame.} =
{.emit: """
var escaped = `b`.replace(/[\-\[\]\/\{\}\(\)\*\+\?\.\\\^\$\|]/g, "\\$&");
return new RegExp("\\b" + escaped + "\\b").test(`a`);
""".}
#proc containsWord(a, b: cstring): bool {.asmNoStackFrame.} =
#{.emit: """
#var escaped = `b`.replace(/[\-\[\]\/\{\}\(\)\*\+\?\.\\\^\$\|]/g, "\\$&");
#return new RegExp("\\b" + escaped + "\\b").test(`a`);
#""".}
proc isWhitespace(text: cstring): bool {.asmNoStackFrame.} =
{.emit: """
@ -252,24 +251,29 @@ proc dosearch(value: cstring): Element =
`stuff` = doc.documentElement;
""".}
db = stuff.getElementsByClass"reference external"
db = stuff.getElementsByClass"reference"
contents = @[]
for ahref in db:
contents.add ahref.textContent.normalize
contents.add ahref.getAttribute("data-doc-search-tag")
let ul = tree("UL")
result = tree("DIV")
result.setClass"search_results"
var matches: seq[(Element, int)] = @[]
let key = value.normalize
for i in 0..<db.len:
let c = contents[i]
if c.containsWord(key):
matches.add((db[i], -(30_000 - c.len)))
elif c.contains(key):
matches.add((db[i], c.len))
if c == "Examples" or c == "PEG construction":
# Some manual exclusions.
# Ideally these should be fixed in the index to be more
# descriptive of what they are.
continue
let (score, matched) = fuzzymatch(value, c)
if matched:
matches.add((db[i], score))
matches.sort do (a, b: auto) -> int:
a[1] - b[1]
for i in 0..min(<matches.len, 19):
b[1] - a[1]
for i in 0 ..< min(matches.len, 19):
matches[i][0].innerHTML = matches[i][0].getAttribute("data-doc-search-tag")
ul.add(tree("LI", matches[i][0]))
if ul.len == 0:
result.add tree("B", text"no search results")