diff options
author | Marius <mariausol@gmail.com> | 2013-05-19 20:40:34 +0300 |
---|---|---|
committer | Marius <mariausol@gmail.com> | 2013-05-19 20:40:34 +0300 |
commit | 13ec4b540e0d46c97fd7b089e0b7413da81e0a9f (patch) | |
tree | bebfa563a17c06b3bd3bf8f6f4ba6d025e00d107 /tex/context/base/util-sql-imp-client.lua | |
parent | 69ad13650cda027526271179e95b5294694143a1 (diff) | |
download | context-13ec4b540e0d46c97fd7b089e0b7413da81e0a9f.tar.gz |
beta 2013.05.19 19:27
Diffstat (limited to 'tex/context/base/util-sql-imp-client.lua')
-rw-r--r-- | tex/context/base/util-sql-imp-client.lua | 512 |
1 files changed, 256 insertions, 256 deletions
diff --git a/tex/context/base/util-sql-imp-client.lua b/tex/context/base/util-sql-imp-client.lua index e09dfde94..7c713a899 100644 --- a/tex/context/base/util-sql-imp-client.lua +++ b/tex/context/base/util-sql-imp-client.lua @@ -1,256 +1,256 @@ -if not modules then modules = { } end modules ['util-sql-client'] = { - version = 1.001, - comment = "companion to util-sql.lua", - author = "Hans Hagen, PRAGMA-ADE, Hasselt NL", - copyright = "PRAGMA ADE / ConTeXt Development Team", - license = "see context related readme files" -} - --- todo: make a converter - -local rawset, setmetatable = rawset, setmetatable -local P, S, V, C, Cs, Ct, Cc, Cg, Cf, patterns, lpegmatch = lpeg.P, lpeg.S, lpeg.V, lpeg.C, lpeg.Cs, lpeg.Ct, lpeg.Cc, lpeg.Cg, lpeg.Cf, lpeg.patterns, lpeg.match - -local trace_sql = false trackers.register("sql.trace", function(v) trace_sql = v end) -local trace_queries = false trackers.register("sql.queries",function(v) trace_queries = v end) -local report_state = logs.reporter("sql","client") - -local sql = utilities.sql -local helpers = sql.helpers -local methods = sql.methods -local validspecification = helpers.validspecification -local preparetemplate = helpers.preparetemplate -local splitdata = helpers.splitdata -local replacetemplate = utilities.templates.replace -local serialize = sql.serialize -local deserialize = sql.deserialize - --- Experiments with an p/action demonstrated that there is not much gain. We could do a runtime --- capture but creating all the small tables is not faster and it doesn't work well anyway. - -local separator = P("\t") -local newline = patterns.newline -local empty = Cc("") - -local entry = C((1-separator-newline)^0) -- C 10% faster than Cs - -local unescaped = P("\\n") / "\n" - + P("\\t") / "\t" - + P("\\0") / "\000" - + P("\\\\") / "\\" - -local entry = Cs((unescaped + (1-separator-newline))^0) -- C 10% faster than Cs but Cs needed due to nesting - -local getfirst = Ct( entry * (separator * (entry+empty))^0) + newline -local skipfirst = (1-newline)^1 * newline -local getfirstline = C((1-newline)^0) - -local cache = { } - -local function splitdata(data) -- todo: hash on first line ... maybe move to client module - if data == "" then - if trace_sql then - report_state("no data") - end - return { }, { } - end - local first = lpegmatch(getfirstline,data) - if not first then - if trace_sql then - report_state("no data") - end - return { }, { } - end - local p = cache[first] - if p then - -- report_state("reusing: %s",first) - local entries = lpegmatch(p.parser,data) - return entries or { }, p.keys - elseif p == false then - return { }, { } - elseif p == nil then - local keys = lpegmatch(getfirst,first) or { } - if #keys == 0 then - if trace_sql then - report_state("no banner") - end - cache[first] = false - return { }, { } - end - -- quite generic, could be a helper - local n = #keys - if n == 0 then - report_state("no fields") - cache[first] = false - return { }, { } - end - if n == 1 then - local key = keys[1] - if trace_sql then - report_state("one field with name %a",key) - end - p = Cg(Cc(key) * entry) - else - for i=1,n do - local key = keys[i] - if trace_sql then - report_state("field %s has name %a",i,key) - end - local s = Cg(Cc(key) * entry) - if p then - p = p * separator * s - else - p = s - end - end - end - p = Cf(Ct("") * p,rawset) * newline^1 - p = skipfirst * Ct(p^0) - cache[first] = { parser = p, keys = keys } - local entries = lpegmatch(p,data) - return entries or { }, keys - end -end - -local splitter = skipfirst * Ct((Ct(entry * (separator * entry)^0) * newline^1)^0) - -local function getdata(data) - return lpegmatch(splitter,data) -end - -helpers.splitdata = splitdata -helpers.getdata = getdata - -local function dataprepared(specification) - local query = preparetemplate(specification) - if query then - io.savedata(specification.queryfile,query) - os.remove(specification.resultfile) - if trace_queries then - report_state("query: %s",query) - end - return true - else - -- maybe push an error - os.remove(specification.queryfile) - os.remove(specification.resultfile) - end -end - -local function datafetched(specification,runner) - local command = replacetemplate(runner,specification) - if trace_sql then - local t = osclock() - report_state("command: %s",command) - local okay = os.execute(command) - report_state("fetchtime: %.3f sec",osclock()-t) -- not okay under linux - return okay == 0 - else - return os.execute(command) == 0 - end -end - -local function dataloaded(specification) - if trace_sql then - local t = osclock() - local data = io.loaddata(specification.resultfile) or "" - report_state("datasize: %.3f MB",#data/1024/1024) - report_state("loadtime: %.3f sec",osclock()-t) - return data - else - return io.loaddata(specification.resultfile) or "" - end -end - -local function dataconverted(data,converter) - if converter then - local data = getdata(data) - if data then - data = converter.client(data) - end - return data - elseif trace_sql then - local t = osclock() - local data, keys = splitdata(data,target) - report_state("converttime: %.3f",osclock()-t) - report_state("keys: %s ",#keys) - report_state("entries: %s ",#data) - return data, keys - else - return splitdata(data) - end -end - --- todo: new, etc - -local function execute(specification) - if trace_sql then - report_state("executing client") - end - if not validspecification(specification) then - report_state("error in specification") - return - end - if not dataprepared(specification) then - report_state("error in preparation") - return - end - if not datafetched(specification,methods.client.runner) then - report_state("error in fetching, query: %s",string.collapsespaces(io.loaddata(specification.queryfile))) - return - end - local data = dataloaded(specification) - if not data then - report_state("error in loading") - return - end - local data, keys = dataconverted(data,specification.converter) - if not data then - report_state("error in converting or no data") - return - end - local one = data[1] - if one then - setmetatable(data,{ __index = one } ) - end - return data, keys -end - --- The following is not that (memory) efficient but normally we will use --- the lib anyway. Of course we could make a dedicated converter and/or --- hook into the splitter code but ... it makes not much sense because then --- we can as well move the builder to the library modules. --- --- Here we reuse data as the indexes are the same, unless we hash. - -local wraptemplate = [[ -local converters = utilities.sql.converters -local deserialize = utilities.sql.deserialize - -local tostring = tostring -local tonumber = tonumber -local booleanstring = string.booleanstring - -%s - -return function(data) - local target = %s -- data or { } - for i=1,#data do - local cells = data[i] - target[%s] = { - %s - } - end - return target -end -]] - -local celltemplate = "cells[%s]" - -methods.client = { - runner = [[mysql --batch --user="%username%" --password="%password%" --host="%host%" --port=%port% --database="%database%" --default-character-set=utf8 < "%queryfile%" > "%resultfile%"]], - execute = execute, - usesfiles = true, - wraptemplate = wraptemplate, - celltemplate = celltemplate, -} +if not modules then modules = { } end modules ['util-sql-client'] = {
+ version = 1.001,
+ comment = "companion to util-sql.lua",
+ author = "Hans Hagen, PRAGMA-ADE, Hasselt NL",
+ copyright = "PRAGMA ADE / ConTeXt Development Team",
+ license = "see context related readme files"
+}
+
+-- todo: make a converter
+
+local rawset, setmetatable = rawset, setmetatable
+local P, S, V, C, Cs, Ct, Cc, Cg, Cf, patterns, lpegmatch = lpeg.P, lpeg.S, lpeg.V, lpeg.C, lpeg.Cs, lpeg.Ct, lpeg.Cc, lpeg.Cg, lpeg.Cf, lpeg.patterns, lpeg.match
+
+local trace_sql = false trackers.register("sql.trace", function(v) trace_sql = v end)
+local trace_queries = false trackers.register("sql.queries",function(v) trace_queries = v end)
+local report_state = logs.reporter("sql","client")
+
+local sql = utilities.sql
+local helpers = sql.helpers
+local methods = sql.methods
+local validspecification = helpers.validspecification
+local preparetemplate = helpers.preparetemplate
+local splitdata = helpers.splitdata
+local replacetemplate = utilities.templates.replace
+local serialize = sql.serialize
+local deserialize = sql.deserialize
+
+-- Experiments with an p/action demonstrated that there is not much gain. We could do a runtime
+-- capture but creating all the small tables is not faster and it doesn't work well anyway.
+
+local separator = P("\t")
+local newline = patterns.newline
+local empty = Cc("")
+
+local entry = C((1-separator-newline)^0) -- C 10% faster than Cs
+
+local unescaped = P("\\n") / "\n"
+ + P("\\t") / "\t"
+ + P("\\0") / "\000"
+ + P("\\\\") / "\\"
+
+local entry = Cs((unescaped + (1-separator-newline))^0) -- C 10% faster than Cs but Cs needed due to nesting
+
+local getfirst = Ct( entry * (separator * (entry+empty))^0) + newline
+local skipfirst = (1-newline)^1 * newline
+local getfirstline = C((1-newline)^0)
+
+local cache = { }
+
+local function splitdata(data) -- todo: hash on first line ... maybe move to client module
+ if data == "" then
+ if trace_sql then
+ report_state("no data")
+ end
+ return { }, { }
+ end
+ local first = lpegmatch(getfirstline,data)
+ if not first then
+ if trace_sql then
+ report_state("no data")
+ end
+ return { }, { }
+ end
+ local p = cache[first]
+ if p then
+ -- report_state("reusing: %s",first)
+ local entries = lpegmatch(p.parser,data)
+ return entries or { }, p.keys
+ elseif p == false then
+ return { }, { }
+ elseif p == nil then
+ local keys = lpegmatch(getfirst,first) or { }
+ if #keys == 0 then
+ if trace_sql then
+ report_state("no banner")
+ end
+ cache[first] = false
+ return { }, { }
+ end
+ -- quite generic, could be a helper
+ local n = #keys
+ if n == 0 then
+ report_state("no fields")
+ cache[first] = false
+ return { }, { }
+ end
+ if n == 1 then
+ local key = keys[1]
+ if trace_sql then
+ report_state("one field with name %a",key)
+ end
+ p = Cg(Cc(key) * entry)
+ else
+ for i=1,n do
+ local key = keys[i]
+ if trace_sql then
+ report_state("field %s has name %a",i,key)
+ end
+ local s = Cg(Cc(key) * entry)
+ if p then
+ p = p * separator * s
+ else
+ p = s
+ end
+ end
+ end
+ p = Cf(Ct("") * p,rawset) * newline^1
+ p = skipfirst * Ct(p^0)
+ cache[first] = { parser = p, keys = keys }
+ local entries = lpegmatch(p,data)
+ return entries or { }, keys
+ end
+end
+
+local splitter = skipfirst * Ct((Ct(entry * (separator * entry)^0) * newline^1)^0)
+
+local function getdata(data)
+ return lpegmatch(splitter,data)
+end
+
+helpers.splitdata = splitdata
+helpers.getdata = getdata
+
+local function dataprepared(specification)
+ local query = preparetemplate(specification)
+ if query then
+ io.savedata(specification.queryfile,query)
+ os.remove(specification.resultfile)
+ if trace_queries then
+ report_state("query: %s",query)
+ end
+ return true
+ else
+ -- maybe push an error
+ os.remove(specification.queryfile)
+ os.remove(specification.resultfile)
+ end
+end
+
+local function datafetched(specification,runner)
+ local command = replacetemplate(runner,specification)
+ if trace_sql then
+ local t = osclock()
+ report_state("command: %s",command)
+ local okay = os.execute(command)
+ report_state("fetchtime: %.3f sec",osclock()-t) -- not okay under linux
+ return okay == 0
+ else
+ return os.execute(command) == 0
+ end
+end
+
+local function dataloaded(specification)
+ if trace_sql then
+ local t = osclock()
+ local data = io.loaddata(specification.resultfile) or ""
+ report_state("datasize: %.3f MB",#data/1024/1024)
+ report_state("loadtime: %.3f sec",osclock()-t)
+ return data
+ else
+ return io.loaddata(specification.resultfile) or ""
+ end
+end
+
+local function dataconverted(data,converter)
+ if converter then
+ local data = getdata(data)
+ if data then
+ data = converter.client(data)
+ end
+ return data
+ elseif trace_sql then
+ local t = osclock()
+ local data, keys = splitdata(data,target)
+ report_state("converttime: %.3f",osclock()-t)
+ report_state("keys: %s ",#keys)
+ report_state("entries: %s ",#data)
+ return data, keys
+ else
+ return splitdata(data)
+ end
+end
+
+-- todo: new, etc
+
+local function execute(specification)
+ if trace_sql then
+ report_state("executing client")
+ end
+ if not validspecification(specification) then
+ report_state("error in specification")
+ return
+ end
+ if not dataprepared(specification) then
+ report_state("error in preparation")
+ return
+ end
+ if not datafetched(specification,methods.client.runner) then
+ report_state("error in fetching, query: %s",string.collapsespaces(io.loaddata(specification.queryfile)))
+ return
+ end
+ local data = dataloaded(specification)
+ if not data then
+ report_state("error in loading")
+ return
+ end
+ local data, keys = dataconverted(data,specification.converter)
+ if not data then
+ report_state("error in converting or no data")
+ return
+ end
+ local one = data[1]
+ if one then
+ setmetatable(data,{ __index = one } )
+ end
+ return data, keys
+end
+
+-- The following is not that (memory) efficient but normally we will use
+-- the lib anyway. Of course we could make a dedicated converter and/or
+-- hook into the splitter code but ... it makes not much sense because then
+-- we can as well move the builder to the library modules.
+--
+-- Here we reuse data as the indexes are the same, unless we hash.
+
+local wraptemplate = [[
+local converters = utilities.sql.converters
+local deserialize = utilities.sql.deserialize
+
+local tostring = tostring
+local tonumber = tonumber
+local booleanstring = string.booleanstring
+
+%s
+
+return function(data)
+ local target = %s -- data or { }
+ for i=1,#data do
+ local cells = data[i]
+ target[%s] = {
+ %s
+ }
+ end
+ return target
+end
+]]
+
+local celltemplate = "cells[%s]"
+
+methods.client = {
+ runner = [[mysql --batch --user="%username%" --password="%password%" --host="%host%" --port=%port% --database="%database%" --default-character-set=utf8 < "%queryfile%" > "%resultfile%"]],
+ execute = execute,
+ usesfiles = true,
+ wraptemplate = wraptemplate,
+ celltemplate = celltemplate,
+}
|