cgit-about-filter: share column widths between tablesHEADmaster

1 files changed, 90 insertions, 0 deletions

diff --git a/pkgs/simple/cgit-about-filter/cgit-about-filter.lua b/pkgs/simple/cgit-about-filter/cgit-about-filter.lua
index d4f441a..cd20d54 100644
--- a/pkgs/simple/cgit-about-filter/cgit-about-filter.lua
+++ b/pkgs/simple/cgit-about-filter/cgit-about-filter.lua
@@ -96,6 +96,95 @@ div#cgit div#summary td + td {
 </style>
 ]]
 
+-- Column widths, shared by all tables with the same header.  A column's
+-- weight is the larger of its longest word and its average cell length, in
+-- characters; code counts more, as monospace is wider.  Narrow columns get a
+-- fixed width, the others a share of what's left.
+local code_weight = 1.25
+local narrow = 12
+
+local function length(s)
+    s = s:gsub("&#?%w+;", "x")
+    return select(2, s:gsub("[^\128-\191]", ""))
+end
+
+local function cell_weights(cell)
+    local longest, total = 0, 0
+    local function add(text, factor)
+        for word in text:gmatch("%S+") do
+            local n = length(word) * factor
+            longest = math.max(longest, n)
+            total = total + n + 1
+        end
+    end
+    for code in cell:gmatch("<code>(.-)</code>") do
+        add((code:gsub("<[^>]+>", "")), code_weight)
+    end
+    add((cell:gsub("<code>.-</code>", " "):gsub("<[^>]+>", " ")), 1)
+    return longest, math.max(total - 1, 0)
+end
+
+local function cells(row)
+    local result = {}
+    for cell in row:gmatch("<t[hd][^>]*>(.-)</t[hd]>") do
+        table.insert(result, cell)
+    end
+    return result
+end
+
+local function header(tbl)
+    return table.concat(cells(tbl:match("<tr>(.-)</tr>") or ""), "\0")
+end
+
+local function column_widths(output)
+    local stats = {}
+    for tbl in output:gmatch("<table>.-</table>") do
+        local key = header(tbl)
+        stats[key] = stats[key] or { rows = 0, longest = {}, total = {} }
+        local s = stats[key]
+        for row in tbl:gmatch("<tr>(.-)</tr>") do
+            s.rows = s.rows + 1
+            for i, cell in ipairs(cells(row)) do
+                local longest, total = cell_weights(cell)
+                s.longest[i] = math.max(s.longest[i] or 0, longest)
+                s.total[i] = (s.total[i] or 0) + total
+            end
+        end
+    end
+    local widths = {}
+    for key, s in pairs(stats) do
+        local weights, wide = {}, 0
+        for i = 1, #s.longest do
+            weights[i] = math.max(s.longest[i], s.total[i] / s.rows)
+            if weights[i] > narrow then
+                wide = wide + weights[i]
+            end
+        end
+        local cols = {}
+        for i, weight in ipairs(weights) do
+            if weight > narrow then
+                cols[i] = string.format('<col style="width:%.1f%%">', 100 * weight / wide)
+            else
+                cols[i] = string.format('<col style="width:calc(%dch + 1.2em)">', math.ceil(weight) + 1)
+            end
+        end
+        widths[key] = "<colgroup>" .. table.concat(cols) .. "</colgroup>"
+    end
+    return widths
+end
+
+-- Add the column widths, and let code break after underscores.
+local function layout_tables(output)
+    local widths = column_widths(output)
+    return (output:gsub("<table>.-</table>", function(tbl)
+        local key = header(tbl)
+        tbl = tbl:gsub("<code>(.-)</code>", function(code)
+            return "<code>" .. code:gsub("_", "_<wbr>") .. "</code>"
+        end)
+        return "<table>" .. widths[key] .. tbl:sub(#"<table>" + 1)
+    end))
+end
+
 local buffer = ""
 local should_render = true
 local current_dir = "" -- Stores the directory prefix of the file being processed
@@ -169,6 +258,7 @@ function filter_close()
     if output then
         if output:find("<table>", 1, true) then
             emit(table_style)
+            output = layout_tables(output)
         end
         emit(output)
     else