Module:parse utilities: Difference between revisions

Line 538:

local split_runs = export.split_alternating_runs_escaping(runs, splitchar, preserve_splitchar, escape_fun,

unescape_fun)

for i = 1, #split_runs~~, (preserve_splitchar and 2 or 1)~~ do

for i = 1, #split_runs do

split_runs[i] = concat(split_runs[i])

end

Line 634:

Parse a term that may have a language code (or possibly multiple plus-separated language codes, if

`data.allow_multiple` is given) preceding it (e.g. {la:minūtia} or {grc:[[σκῶρ|σκατός]]} or

{nan-hbl+hak:[[毋]][[知]]}). Return ~~four~~ arguments:

{nan-hbl+hak:[[毋]][[知]]}). Return five arguments:

# the term ~~minus~~ the ~~language code~~;

# the original prefixed term; in the case of a Wikipedia or Wikisource prefix followed by a two-part link, it is a

two-part link with the Wikipedia/Wikisource prefix moved inside the link; in the case of a Wikipedia or Wikisource

prefix followed by a redundant one-part link, the brackets are removed;

# the language object corresponding to the language code (possibly a family object if `data.allow_family` is given), or

a list of such objects if `data.allow_multiple` is given;

# the link if the term is of the form {[[<var>link</var>|<var>display</var>]]~~} (it may be generated into that form with~~

# the link if the unprefixed term is of the form <code>[[<var>link</var>|<var>display</var>]]</code> or of the form

~~Wikipedia and Wikisource prefixes)~~ or of the form {{[[<var>link</var>]]}, otherwise the full term;

<code>[[<var>link</var>]]</code>, otherwise the full unprefixed term;

# the display part if the term is of the form {[[<var>link</var>|<var>display</var>]]}, ~~otherwise nil~~.

# the display part if the term is of the form <code>[[<var>link</var>|<var>display</var>]]</code> or has a Wikipedia or

Wikisource prefix (in which case the part minus the prefix and any following language code will be returned, with

redundant brackets stripped), else {nil};

# {true} if the term has a Wikipedia/Wikisource prefix, else {false}.

Etymology-only languages are always allowed. This function also correctly handles Wikipedia prefixes (e.g.

{w:Abatemarco} or {w:it:Colle Val d'Elsa} or {lw:ru:Филарет}) and Wikisource prefixes (e.g. {s:Twelve O'Clock} or

Line 697:

Line 702:

local lang = wiki_links and get_lang(foreign_wiki, parse_err, "allow etym") or nil

local prefixed_link = wiki_prefix .. link

return ("[[%s|%s]]"):format(prefixed_link, display or link), lang, prefixed_link, ~~display~~

if display then

return ("[[%s|%s]]"):format(prefixed_link, display), lang, prefixed_link, display, true

else

-- Return the link minus any language codes as the fourth term (display form). Previously we returned `actual_term`

-- but this causes problems with redundant Wikipedia links of the form `w:[[Dragon Ball Z]]`. Don't generate a

-- two-part link so you can specify a display form in 3=. Note that the fourth and fifth params are currently only

-- used in [[Module:quote]].

return prefixed_link, lang, prefixed_link, link, true

end

Line 792:

Line 805:

end

local link, display = parse_bracketed_term(term, parse_err)

return term, termlang, link, display

return term, termlang, link, display, false

end

Line 1,102:

Line 1,115:

if props.param_mods[prefix] then

local mod_props = props.param_mods[prefix]

if mod_props.replaced_by == false then

prefix_parse_err(

("Prefix has been removed and is no longer valid%s%s"):format(

mod_props.reason and ", " .. mod_props.reason or "",

mod_props.instead and "; instead, " .. mod_props.instead or "")

)

elseif mod_props.replaced_by then

prefix_parse_err(

("Prefix has been replaced by '%s'%s"):format(

mod_props.replaced_by, mod_props.reason and ", " .. mod_props.reason or "")

)

end

local key = mod_props.item_dest or prefix

local dest

@@ Line 538: / Line 538: @@
 			local split_runs = export.split_alternating_runs_escaping(runs, splitchar, preserve_splitchar, escape_fun,
 				unescape_fun)
-			for i = 1, #split_runs, (preserve_splitchar and 2 or 1) do
+			for i = 1, #split_runs do
 				split_runs[i] = concat(split_runs[i])
 			end
@@ Line 634: / Line 634: @@
 Parse a term that may have a language code (or possibly multiple plus-separated language codes, if
 `data.allow_multiple` is given) preceding it (e.g. {la:minūtia} or {grc:[[σκῶρ|σκατός]]} or
-{nan-hbl+hak:[[毋]][[知]]}). Return four arguments:
+{nan-hbl+hak:[[毋]][[知]]}). Return five arguments:
-# the term minus the language code;
+# the original prefixed term; in the case of a Wikipedia or Wikisource prefix followed by a two-part link, it is a
+  two-part link with the Wikipedia/Wikisource prefix moved inside the link; in the case of a Wikipedia or Wikisource
+  prefix followed by a redundant one-part link, the brackets are removed;
 # the language object corresponding to the language code (possibly a family object if `data.allow_family` is given), or
    a list of such objects if `data.allow_multiple` is given;
-# the link if the term is of the form {[[<var>link</var>|<var>display</var>]]} (it may be generated into that form with
+# the link if the unprefixed term is of the form <code>[[<var>link</var>|<var>display</var>]]</code> or of the form
-  Wikipedia and Wikisource prefixes) or of the form {{[[<var>link</var>]]}, otherwise the full term;
+  <code>[[<var>link</var>]]</code>, otherwise the full unprefixed term;
-# the display part if the term is of the form {[[<var>link</var>|<var>display</var>]]}, otherwise nil.
+# the display part if the term is of the form <code>[[<var>link</var>|<var>display</var>]]</code> or has a Wikipedia or
+  Wikisource prefix (in which case the part minus the prefix and any following language code will be returned, with
+  redundant brackets stripped), else {nil};
+# {true} if the term has a Wikipedia/Wikisource prefix, else {false}.
 Etymology-only languages are always allowed. This function also correctly handles Wikipedia prefixes (e.g.
 {w:Abatemarco} or {w:it:Colle Val d'Elsa} or {lw:ru:Филарет}) and Wikisource prefixes (e.g. {s:Twelve O'Clock} or
@@ Line 697: / Line 702: @@
 		local lang = wiki_links and get_lang(foreign_wiki, parse_err, "allow etym") or nil
 		local prefixed_link = wiki_prefix .. link
-		return ("[[%s|%s]]"):format(prefixed_link, display or link), lang, prefixed_link, display
+		if display then
+			return ("[[%s|%s]]"):format(prefixed_link, display), lang, prefixed_link, display, true
+		else
+			-- Return the link minus any language codes as the fourth term (display form). Previously we returned `actual_term`
+			-- but this causes problems with redundant Wikipedia links of the form `w:[[Dragon Ball Z]]`. Don't generate a
+			-- two-part link so you can specify a display form in 3=. Note that the fourth and fifth params are currently only
+			-- used in [[Module:quote]].
+			return prefixed_link, lang, prefixed_link, link, true
+		end
 	end
@@ Line 792: / Line 805: @@
 	end
 	local link, display = parse_bracketed_term(term, parse_err)
-	return term, termlang, link, display
+	return term, termlang, link, display, false
 end
@@ Line 1,102: / Line 1,115: @@
 		if props.param_mods[prefix] then
 			local mod_props = props.param_mods[prefix]
+			if mod_props.replaced_by == false then
+				prefix_parse_err(
+					("Prefix has been removed and is no longer valid%s%s"):format(
+						mod_props.reason and ", " .. mod_props.reason or "",
+						mod_props.instead and "; instead, " .. mod_props.instead or "")
+				)
+			elseif mod_props.replaced_by then
+				prefix_parse_err(
+					("Prefix has been replaced by '%s'%s"):format(
+						mod_props.replaced_by, mod_props.reason and ", " .. mod_props.reason or "")
+				)
+			end
 			local key = mod_props.item_dest or prefix
 			local dest