We propose a blog search engine called "Kawaii Search" (where Kawaii means pretty) to search blogs based on the impression of their text on a printing surface, considering factors such as the format and layout of text and density of words. Particularly in Japan, blogs reveal the personality characteristics of users depending on how they place their text. For example, some writers leave more space between lines or use hieroglyphics and "Gal words," which consist of slang or abbreviations. Further, words can be categorized using four types of characters: kanji, hiragana, katakana, and alphabet. Each results in a different impression that reveals a writer's personality. Given this approach, blog readers can not only read blog, but also interpret each writer's personality. By focusing on impression differences, we propose a new search algorithm specialized for Japanese blogs. To show that these differences can act as the base of our search algorithm, we conducted an experiment that successfully verified the algorithm applied to the following three blog patterns: "kawaii" (pretty or lovely), "majime" (seriousness or industrious), and "futsu" (normal). The results show that in terms of the accuracy of the algorithm, our study categorized "kawaii" well; however, "majime" and "futsu" did not show good results.