Programming Forums > Script Programming > Python

Sane · Apr 1st, 2006, 8:19 PM

I'm wondering when the line should be drawn between the speed of producing the output /or/ recalling cached data.

I have a function that will parse limited BBCode to HTML...

(Toggle Plain Text)

    def parse_post(self, text):
        for s in self.smilies:
            text = text.replace(s[0], '<img class="inlineimg" border="0" alt="%s" title="%s" src="%s" />'%(s[0], s[0], s[1]))

        def check_quote(text):
            n = 0
            qs = []
            o = False
            for i in range(len(text)):
                if text[i:i+7] == 'quote="':
                    o = i

                elif text[i] == '"' and o and i-o > 7:
                    n += 1
                    qs.append(text[o+7:i])
                    o = False
                    
                elif text[i:i+8] == "[/quote]":
                    n -= 1

                if n < 0:
                    return False

            if qs:
                return qs
            
            else:
                return n == 0

        def check_tag(x, y, bal=False):
            n = 0
            leny = len(y)
            ts = [ '[%s]'%y, '[/%s]'%y ]
            
            for i in range(len(x)-leny-2):
                if x[i:i+leny+2] == ts[0]:
                    n += 1
                elif x[i:i+leny+3] == ts[1]:
                    n -= 1

                if n < 0:
                    return False
                if bal and n > 1:
                    return False

            return n == 0
        
        def get_between(x, s, e):
            try:
                return x.split(s)[1].split(e)[0]
            except IndexError: return False
            
        warning = ''
        g = ['b', 'u', 'i']

        for gg in g:
            if check_tag(text, gg):
                text = text.replace('[%s]'%gg, '<%s>'%gg)
                text = text.replace('[/%s]'%gg, '</%s>'%gg)
            else:
                warning += "<b>Warning</b> Your '<i>%s</i>' tag(s) were not formatted correctly.<br />"%gg

        if check_tag(text, 'img', True):
            text = text.replace('[img]', '<img src="')
            text = text.replace('[/img]', '" />')
        else:
            warning += "<b>Warning</b> Your '<i>IMG</i>' tag(s) were not formatted correctly.<br />"
            
        if check_tag(text, 'url', True):
            while 1:
                lnk = get_between(text, '', '')
                if not lnk:
                    break
                else:
                    text = text.replace('%s'%lnk, '<a href="%s">%s</a>'%(lnk, lnk), 1)
        else:
            warning += "<b>Warning</b> Your '<i>URL</i>' tag(s) were not formatted correctly.<br />"

        names = check_quote(text)
        if type(names) == list:
            for name in names:
                text = text.replace( '[INSERT WORD QUOTE HERE="%s"]'%name, '<br /><br /> &nbsp; &nbsp; Quote by <b>%s</b><br /><div class="quote">'%name )
            text = text.replace( '[/quote]', '</div><br /><br />' )
        else:
            if not names:
                warning += "<b>Warning</b> Your '<i>quote</i>' tag(s) were not formatted correctly.<br />"
        
        return text+['','<br /><br /><hr />'+warning][len(warning)>0]

    def parse_post(self, text):
        for s in self.smilies:
            text = text.replace(s[0], '<img class="inlineimg" border="0" alt="%s" title="%s" src="%s" />'%(s[0], s[0], s[1]))

        def check_quote(text):
            n = 0
            qs = []
            o = False
            for i in range(len(text)):
                if text[i:i+7] == 'quote="':
                    o = i

                elif text[i] == '"' and o and i-o > 7:
                    n += 1
                    qs.append(text[o+7:i])
                    o = False
                    
                elif text[i:i+8] == "[/quote]":
                    n -= 1

                if n < 0:
                    return False

            if qs:
                return qs
            
            else:
                return n == 0

        def check_tag(x, y, bal=False):
            n = 0
            leny = len(y)
            ts = [ '[%s]'%y, '[/%s]'%y ]
            
            for i in range(len(x)-leny-2):
                if x[i:i+leny+2] == ts[0]:
                    n += 1
                elif x[i:i+leny+3] == ts[1]:
                    n -= 1

                if n < 0:
                    return False
                if bal and n > 1:
                    return False

            return n == 0
        
        def get_between(x, s, e):
            try:
                return x.split(s)[1].split(e)[0]
            except IndexError: return False
            
        warning = ''
        g = ['b', 'u', 'i']

        for gg in g:
            if check_tag(text, gg):
                text = text.replace('[%s]'%gg, '<%s>'%gg)
                text = text.replace('[/%s]'%gg, '</%s>'%gg)
            else:
                warning += "<b>Warning</b> Your '<i>%s</i>' tag(s) were not formatted correctly.<br />"%gg

        if check_tag(text, 'img', True):
            text = text.replace('[img]', '<img src="')
            text = text.replace('[/img]', '" />')
        else:
            warning += "<b>Warning</b> Your '<i>IMG</i>' tag(s) were not formatted correctly.<br />"
            
        if check_tag(text, 'url', True):
            while 1:
                lnk = get_between(text, '', '')
                if not lnk:
                    break
                else:
                    text = text.replace('%s'%lnk, '<a href="%s">%s</a>'%(lnk, lnk), 1)
        else:
            warning += "<b>Warning</b> Your '<i>URL</i>' tag(s) were not formatted correctly.<br />"

        names = check_quote(text)
        if type(names) == list:
            for name in names:
                text = text.replace( '[INSERT WORD QUOTE HERE="%s"]'%name, '<br /><br /> &nbsp; &nbsp; Quote by <b>%s</b><br /><div class="quote">'%name )
            text = text.replace( '[/quote]', '</div><br /><br />' )
        else:
            if not names:
                warning += "<b>Warning</b> Your '<i>quote</i>' tag(s) were not formatted correctly.<br />"
        
        return text+['','<br /><br /><hr />'+warning][len(warning)>0]

Will it be faster to cache the results and recall that result, or just calculate it again each time? My RAM is 512 and it's only running one python process.

Edit: The source contains INSERT WORD QUOTE HERE, since having the word 'quote' plainly there made quote tags appear. Gah, the BBCode is driving my source nuts in some places. Just quote my post to see the unparsed text.

Arevos · Apr 2nd, 2006, 5:56 AM

It's usually always faster to cache, especially since Python's dictionaries are generally pretty efficient.

Usually when dealing with text, the original text isn't used as the key, however. A MD5 or SHA1 hash is used instead to save memory and key-finding times. So instead of having a caching dictionary like:

(Toggle Plain Text)

{ original_text : converted_text }

Use a dictionary like:

(Toggle Plain Text)

{ sha.new(original_text).hexdigest() : converted_text }

Since you're potentially dealing with large chunks of text, you might also want to put a limit on your cache of some kind. Maybe cached entries are only around for a fixed amount of time, or maybe the dictionary is assigned a fixed amount of memory, and the oldest entry is deleted to make way for the newest.

Also, if you're parsing, you might want to check out PyParsing. It's a lot easier than doing things by hand.

Sane · Apr 2nd, 2006, 12:19 PM

(Toggle Plain Text)

{ sha.new(original_text).hexdigest() : converted_text }

That's brilliant! Wow, that's a really good idea.

Dameon · Apr 2nd, 2006, 12:25 PM

Some more info regarding usage of this program would help in that decision. Is it running on a webserver? If so, is the cache even available across multiple requests? How long is the typical block of BBCode tags? And most importantly: How often does original text repeat? Why cache what is never called up?

Sane · Apr 2nd, 2006, 3:10 PM

Well before it was reparsing every post every time someone visits a thread. This is because I want to parse everything after it's saved to the data files (in case somebody finds a glitch in the parser, or I want to make changes to the effects of the tags).

But now it only needs to parse each post once, until the webserver is rebooted. It's working fine. Instead of calling self.parse_post(), I just changed that to self.gateway_parse()

(Toggle Plain Text)

    def gateway_parse(self, post):
        checksum = sha(post).hexdigest()
        try:
            res = self.cache_parse[checksum]
        except KeyError:
            res = self.parse_post(post)
            self.cache_parse[checksum] = res
        return res

Arevos · Apr 2nd, 2006, 4:11 PM

Quote:

Originally Posted by Sane

That's brilliant! Wow, that's a really good idea.

That's more or less what I said when I was introduced to that method by a programmer on the Drupal mailing list some years ago

Apr 1st, 2006, 8:19 PM	#1
Sane Programming Guru Join Date: Apr 2005 Location: Waterloo, Ontario Posts: 2,086 Rep Power: 6	When is it faster to cache? I'm wondering when the line should be drawn between the speed of producing the output /or/ recalling cached data. I have a function that will parse limited BBCode to HTML... (Toggle Plain Text) def parse_post(self, text): for s in self.smilies: text = text.replace(s[0], '<img class="inlineimg" border="0" alt="%s" title="%s" src="%s" />'%(s[0], s[0], s[1])) def check_quote(text): n = 0 qs = [] o = False for i in range(len(text)): if text[i:i+7] == 'quote="': o = i elif text[i] == '"' and o and i-o > 7: n += 1 qs.append(text[o+7:i]) o = False elif text[i:i+8] == "[/quote]": n -= 1 if n < 0: return False if qs: return qs else: return n == 0 def check_tag(x, y, bal=False): n = 0 leny = len(y) ts = [ '[%s]'%y, '[/%s]'%y ] for i in range(len(x)-leny-2): if x[i:i+leny+2] == ts[0]: n += 1 elif x[i:i+leny+3] == ts[1]: n -= 1 if n < 0: return False if bal and n > 1: return False return n == 0 def get_between(x, s, e): try: return x.split(s)[1].split(e)[0] except IndexError: return False warning = '' g = ['b', 'u', 'i'] for gg in g: if check_tag(text, gg): text = text.replace('[%s]'%gg, '<%s>'%gg) text = text.replace('[/%s]'%gg, '</%s>'%gg) else: warning += "<b>Warning</b> Your '<i>%s</i>' tag(s) were not formatted correctly.<br />"%gg if check_tag(text, 'img', True): text = text.replace('[img]', '<img src="') text = text.replace('[/img]', '" />') else: warning += "<b>Warning</b> Your '<i>IMG</i>' tag(s) were not formatted correctly.<br />" if check_tag(text, 'url', True): while 1: lnk = get_between(text, '', '') if not lnk: break else: text = text.replace('%s'%lnk, '<a href="%s">%s</a>'%(lnk, lnk), 1) else: warning += "<b>Warning</b> Your '<i>URL</i>' tag(s) were not formatted correctly.<br />" names = check_quote(text) if type(names) == list: for name in names: text = text.replace( '[INSERT WORD QUOTE HERE="%s"]'%name, '<br /><br />     Quote by <b>%s</b><br /><div class="quote">'%name ) text = text.replace( '[/quote]', '</div><br /><br />' ) else: if not names: warning += "<b>Warning</b> Your '<i>quote</i>' tag(s) were not formatted correctly.<br />" return text+['','<br /><br /><hr />'+warning][len(warning)>0] def parse_post(self, text): for s in self.smilies: text = text.replace(s[0], '<img class="inlineimg" border="0" alt="%s" title="%s" src="%s" />'%(s[0], s[0], s[1])) def check_quote(text): n = 0 qs = [] o = False for i in range(len(text)): if text[i:i+7] == 'quote="': o = i elif text[i] == '"' and o and i-o > 7: n += 1 qs.append(text[o+7:i]) o = False elif text[i:i+8] == "[/quote]": n -= 1 if n < 0: return False if qs: return qs else: return n == 0 def check_tag(x, y, bal=False): n = 0 leny = len(y) ts = [ '[%s]'%y, '[/%s]'%y ] for i in range(len(x)-leny-2): if x[i:i+leny+2] == ts[0]: n += 1 elif x[i:i+leny+3] == ts[1]: n -= 1 if n < 0: return False if bal and n > 1: return False return n == 0 def get_between(x, s, e): try: return x.split(s)[1].split(e)[0] except IndexError: return False warning = '' g = ['b', 'u', 'i'] for gg in g: if check_tag(text, gg): text = text.replace('[%s]'%gg, '<%s>'%gg) text = text.replace('[/%s]'%gg, '</%s>'%gg) else: warning += "<b>Warning</b> Your '<i>%s</i>' tag(s) were not formatted correctly.<br />"%gg if check_tag(text, 'img', True): text = text.replace('[img]', '<img src="') text = text.replace('[/img]', '" />') else: warning += "<b>Warning</b> Your '<i>IMG</i>' tag(s) were not formatted correctly.<br />" if check_tag(text, 'url', True): while 1: lnk = get_between(text, '', '') if not lnk: break else: text = text.replace('%s'%lnk, '<a href="%s">%s</a>'%(lnk, lnk), 1) else: warning += "<b>Warning</b> Your '<i>URL</i>' tag(s) were not formatted correctly.<br />" names = check_quote(text) if type(names) == list: for name in names: text = text.replace( '[INSERT WORD QUOTE HERE="%s"]'%name, '<br /><br />     Quote by <b>%s</b><br /><div class="quote">'%name ) text = text.replace( '[/quote]', '</div><br /><br />' ) else: if not names: warning += "<b>Warning</b> Your '<i>quote</i>' tag(s) were not formatted correctly.<br />" return text+['','<br /><br /><hr />'+warning][len(warning)>0] Will it be faster to cache the results and recall that result, or just calculate it again each time? My RAM is 512 and it's only running one python process. Edit: The source contains INSERT WORD QUOTE HERE, since having the word 'quote' plainly there made quote tags appear. Gah, the BBCode is driving my source nuts in some places. Just quote my post to see the unparsed text.

Apr 2nd, 2006, 5:56 AM	#2
Arevos Programming Guru Join Date: Aug 2005 Location: England Posts: 1,499 Rep Power: 5	It's usually always faster to cache, especially since Python's dictionaries are generally pretty efficient. Usually when dealing with text, the original text isn't used as the key, however. A MD5 or SHA1 hash is used instead to save memory and key-finding times. So instead of having a caching dictionary like: (Toggle Plain Text) { original_text : converted_text } { original_text : converted_text } Use a dictionary like: (Toggle Plain Text) { sha.new(original_text).hexdigest() : converted_text } { sha.new(original_text).hexdigest() : converted_text } Since you're potentially dealing with large chunks of text, you might also want to put a limit on your cache of some kind. Maybe cached entries are only around for a fixed amount of time, or maybe the dictionary is assigned a fixed amount of memory, and the oldest entry is deleted to make way for the newest. Also, if you're parsing, you might want to check out PyParsing. It's a lot easier than doing things by hand.

Apr 2nd, 2006, 12:19 PM	#3
Sane Programming Guru Join Date: Apr 2005 Location: Waterloo, Ontario Posts: 2,086 Rep Power: 6	(Toggle Plain Text) { sha.new(original_text).hexdigest() : converted_text } { sha.new(original_text).hexdigest() : converted_text } That's brilliant! Wow, that's a really good idea.

Apr 2nd, 2006, 12:25 PM	#4
Dameon Troll Join Date: Apr 2005 Location: Texas Posts: 732 Rep Power: 4	Some more info regarding usage of this program would help in that decision. Is it running on a webserver? If so, is the cache even available across multiple requests? How long is the typical block of BBCode tags? And most importantly: How often does original text repeat? Why cache what is never called up? __________________ MD5(sig) = bcef75433db02e9ad9bf81d6f7c5c270

Apr 2nd, 2006, 3:10 PM	#5
Sane Programming Guru Join Date: Apr 2005 Location: Waterloo, Ontario Posts: 2,086 Rep Power: 6	Well before it was reparsing every post every time someone visits a thread. This is because I want to parse everything after it's saved to the data files (in case somebody finds a glitch in the parser, or I want to make changes to the effects of the tags). But now it only needs to parse each post once, until the webserver is rebooted. It's working fine. Instead of calling self.parse_post(), I just changed that to self.gateway_parse() (Toggle Plain Text) def gateway_parse(self, post): checksum = sha(post).hexdigest() try: res = self.cache_parse[checksum] except KeyError: res = self.parse_post(post) self.cache_parse[checksum] = res return res def gateway_parse(self, post): checksum = sha(post).hexdigest() try: res = self.cache_parse[checksum] except KeyError: res = self.parse_post(post) self.cache_parse[checksum] = res return res

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Programming Forums > Script Programming > Python

When is it faster to cache?